Download NBER WORKING PAPER SERIES VOICE AND GROWTH: WAS CHURCHILL RIGHT?

NBER WORKING PAPER SERIES
VOICE AND GROWTH: WAS CHURCHILL RIGHT?
Peter H. Lindert
Working Paper 9749
http://www.nber.org/papers/w9749
NATIONAL BUREAU OF ECONOMIC RESEARCH
1050 Massachusetts Avenue
Cambridge, MA 02138
June 2003
The views expressed herein are those of the authors and not necessarily those of the National Bureau of
Economic Research.
©2003 by Brian Knight. All rights reserved. Short sections of text not to exceed two paragraphs, may be
quoted without explicit permission provided that full credit including © notice, is given to the source.
Voice and Growth: Was Churchill Right?
Peter H. Lindert
NBER Working Paper No. 9749
June 2003
JEL No. H11, N40, O40
ABSTRACT
The debate over whether political democracy is the least bad regime, as Churchill once said,
remains unresolved because history has been ignored or misread, and because recent statistical
studies have not chosen the right tests. Using too little historical information, and mistaking formal
democratic rules for true voice, has understated the gains from spreading political voice more
equally.
This paper draws on a deeper history, reinterpreting five key experiences to show how the
institutional channels linking voice and growth are themselves evolving with the economy. Up to
about the early nineteenth century, the key institutional link was property rights and contract
enforcement. Since the early nineteenth century, the human-investment channel has assumed an
ever-greater role. This trend will probably continue.
A telltale sign of damage to growth from elite rule is the under-investment of public funds
in egalitarian human capital, especially primary schooling, relative to historical norms for successful
economies.
Peter H. Lindert
Department of Economics
University of California, Davis
Davis, CA 95616
and NBER
[email protected]
May 2003
VOICE AND GROWTH:
WAS CHURCHILL RIGHT?
Peter H. Lindert*∗
On the afternoon of November 11th, 1947, the Opposition leader Winston
Churchill gave the House of Commons, and posterity, his famous defense of
democracy:
No one pretends that democracy is perfect or all-wise. Indeed it has been
said that democracy is the worst form of Government except all those other
forms that have been tried from time to time; but there is the broad feeling
in our country that the people should rule, continuously rule, and that public
opinion, expressed by all constitutional means, should shape, guide, and
control the actions of Ministers who are their servants and not their
masters....”1
In fact, Churchill was trying to block the advance of democracy on that
November day. He was defending the power of the House of Lords to block
measures advanced by a popularly elected government. The House of Lords was a
hereditary old boys’ club2 that was not required, and not even allowed, to be elected
by the people. Churchill was in the trenches fighting against a Labour government
bill that would trim the number of years that the Lords could block a bill from two
years to one year. Labour’s purpose in attacking the Lords’ slight remaining powers
was to clear the way for nationalizing the steel industry, a controversial move that
the Lords were determined to block. Our reading of subsequent history arouses at
least some sympathy for Churchill’s opposition to steel nationalization.3 Yet the fact
∗
Professor of Economics, University of California - Davis.
“Voice and Growth,” Page 1
remains that he was fighting for a rich hereditary elite against a popularly elected
government. Churchill lost that battle, and democracy advanced another step.
So far we have two different Churchills on the subject of democracy’s merits
for the economy and civilization: The great orator endorsing popular democracy as
the best we can do, and the conservative defending the last vestiges of hereditary
elite power against the excesses of democracy. Actually, there was at least one other
Churchill on the subject. This third Churchill demanded popular democracy as a
means to achieve a righteous redistribution from landlords to the rest of society. As
a young Liberal rabble-rouser on the election trail in 1909, this younger Churchill
derided the House of Lords as “that home of the ancient British aristocracy.” As he
told a cheering Liberal Club dinner in Birmingham,
The powers of the House of Lords to impede ... are strangely bestowed....
Posing as a chamber of review, remote from popular passion, far from the
swaying influences of the electorate, it nevertheless exhibits ... a party spirit
upon a level with many of the least reputed Chambers in the world.... It is
not possible for reasonable men to defend such a system or such an
institution.4
His reason for poking at the Lords and rallying the mass electorate was
redistributive and revolutionary. Young Churchill was a leading land-tax Liberal, a
temporary disciple of Henry George and an Ally of Lloyd George, demanding
redistribution from landlords to workers and the poor.
To ask “Was Churchill Right?” is therefore to ask: “Which of the three
Churchills was right -- the populist orator, the elitist Conservative, or the young
redistributor?” Let us follow the populist orator, and ask about the economic side of
his saying. Let us take the Churchill view to mean that more popular voice in
democracy is better for economic growth over the long run. Is that a correct reading
of history? Or did China’s government raise its growth chances by crushing
“Voice and Growth,” Page 2
democracy at Tiananmen Square? If superpowers and international agencies want
to promote growth and fight Third World poverty, should they support full
democracy, limited elite democracy, or firm autocracy?
As an aid to renewed exploration of such issues, this paper offers the
following suggestions:
(1) Past attempts to judge the growth impacts of political regimes have used
history too little, have mistaken democratic rules for true voice, and have focused on
an overly narrow concept of property rights. This has led to an under-valuation of
the importance of spreading political voice more equally over the whole population.
(2) Defining democracy less narrowly than in the past literature helps to
reconcile these suggestions with five historical experiences that might have seemed
to suggest negative effects of fuller democracy on economic growth.
(3) The institutional channels that link voice and growth are themselves
evolving with the economy. Up to about the early nineteenth century, the key
institutional link was property rights and contract enforcement. Since the early
nineteenth century, the human-investment policy channel has assumed an evergreater role. This trend will continue.
(4) A telltale sign of damage to growth from elite rule is the under-investment
of public funds in egalitarian human capital, especially primary schooling, relative
to historical norms for successful economies. Policies toward basic human capital
formation are a key to the better growth performance of full-franchise democracies
over either autocracies or (especially) elite democracies.
THE DEBATE OVER DEMOCRACY, INSTITUTIONS, AND GROWTH
Conflicting Views and Defective Tests
“Voice and Growth,” Page 3
Those who have debated the effects of democracy on growth have divided
into four camps: optimists, pessimists, neutralists, and agnostics. The neutralists
and agnostics dominate right now, though I will argue that their dominance is based
on our not having chosen the right empirical measures and the right tests.
Most optimists have shared Churchill’s modesty about the case for
democracy. The twentieth century has left many members, but few Panglossian
Whigs, in this optimistic camp. A fair recent statement of muted optimism is
Douglass North’s 1990 statement that modern democratic society with universal
suffrage is best, as Churchill said, but it’s a poor best.5 The optimist belief that
democracy promotes growth rests on a faith in universal suffrage and participation
as the best way to keep high-level mistakes in check. Somebody has to warn and
threaten the boss, and the more voices that speak out, the better -- yet this may not
suffice to prevent serious mistakes. Optimists share the low expectations of Albert
Hirschman, who argued that every organization, private or public, lapses into bad
performance, and that a mixture of exit and voice is required to minimize the costs
of its errors. Democracy is no cure, but it enhances the voice option, so that the best
critics don’t all exit.6
Pessimists fear that democracy lowers income and growth by yielding to
demagogues and to demands for current consumption, in the form of aid to
declining sectors, power to unions, and taxation for safety nets. Concerning the
Second and Third Worlds, many writers have drawn a political corollary from
Gerschenkron that impoverished countries could awaken and catch up only if
autocrats suppressed consumption and transformed institutions from above, just as
Imperial Germany and Russia did in the past.7
The twentieth century offers pessimists a lot of anecdotal support. Didn’t the
Weimar democracy give Hitler a plurality in 1932? Didn’t fairly democratic Chile
give a plurality to Salvador Allende in 1970, wrecking the economy within three
years? Also gloomy is the more recent behavior of some so-called democracies in
electing Lukashenko in Belarus, Kuchma in Ukraine, and similar economy-wreckers
“Voice and Growth,” Page 4
in other former Soviet republics.8 The mid-1990s were a high tide for fearing the
economic costs of such “illiberal democracies,” and applauding Pinochet, Lee Kuan
Yew, and the other firm hands in control of newly industrializing countries.9
Neutralists feel that the net effect of democracy versus autocracy is zero.
Adam Przeworski and co-authors conclude from a massive statistical study of 137
countries between 1950 and 1990 that
“there is no trade-off between democracy and development, not even in poor
countries.... We hope to have put the entire issue to rest.... [T]he entire
controversy seems to have been much ado about nothing.... [T]he recently
heralded economic virtues of democracy are yet another figment of the
ideological imagination.”10
The neutralist camp includes a series of statistical studies finding that there is no
effect of switching between autocracy and democracy, although regime durability
matters within either type of regime.11 Neutralism also includes those who agree
with Robert Barro and Niall Ferguson that growth is maximized by an intermediate
amount of democracy, beyond which fuller democracy erases any net contribution
to growth.12
Agnostics dismiss all past attempts to measure the growth effects of
democracy as flawed and inconclusive, and with good reasons. Most of the tests fail
to correct sufficiently for simultaneity bias and regime selectivity bias. The survival
of political leaders and political regimes is itself endogenous, along with growth,
education, and economic inequality. It will not do to test for the effects of, for
example, democracy or inequality on the rate of growth without simultaneousequation techniques that incorporate these feedbacks. Regime attrition also affects
the apparent regime-specific rates of growth. Lacking such adjustments, the
discerning reader of the statistical literature remains agnostic about whether
democracy promotes growth.13 The statistical literature on democracy and growth
“Voice and Growth,” Page 5
has failed to meet the requirements for a fair test, for even more reasons than the
agnostics have given. A final section of this paper describes what better tests seem
to show.
Growth versus well-being
For many issues, GDP per capita has proven its usefulness as a strong
correlate of well-being. The issue of democracy and growth is different, however.
Most obviously, freedom is valued for itself, and people in the past have given up
large amounts of income for it.14 The obvious value of freedom itself is set aside
here, to confront the tougher issue of whether income has to be sacrificed for it. Less
dispensable, however, is the effect of political regimes on life expectancy. As the
case of China’s Great Leap has just reminded us, regimes must be judged in part by
the deaths they hasten.
Even within the confines of GDP per capita, growth rates have been
overused. As James Tobin warned when steady-state growth theory had its vogue,
you can’t eat rates. You can only eat levels. In the background here is the deceiving
use of recent income levels as an exogenous control variable. Scholars, including
this author, have used the rates and levels as equivalent in regressions that have
controlled for past income levels. That works well enough for many purposes, but
not if we need to appraise the impact of political regime history, which lasts for
more decades than the income variable covers. The statistical literature on political
regimes and growth hides much of the effect of regime history in the lagged income
variable, as well as in fixed-country effects.
Which concept of democracy?
It is time for scholars in political economy to retreat from a sound scientific
decision they have reached regarding the definition of democracy. To distance their
“Voice and Growth,” Page 6
analysis from any hint of circular reasoning, some writers have insisted that
democracy must be defined only in very narrow procedural terms, as a setting in
which there are formal elections for the chief executive and a legislature, and there is
at least some opposition. Explicitly avoiding any outcome dimensions such as
“accountability,” “responsiveness,” “representation,” “equality,” or “civil rights,” or
any social and economic sources of political privilege has kept this narrow
procedural definition innocent of letting economic success define democracy. The
resulting analysis soundly shows that this narrow definition truly has little or no
effect on economic growth, within the confines of the chosen models and data sets.15
How wide is the audience that will accept this narrow result as the answer to
their curiosity about the economic implications of “democracy,” a concept that has
always meant governance by all adults?16 One can broaden the definition of who
has political voice without plunging down the slippery slope into the circularity of
asserting that democracy is partly defined by the economic success it fosters. We
naturally expect that the greater the share of adults who have political voice, the
more democratic is the polity. Yet scholars’ taxonomies of political regimes miss big
differences in the share of adults who have any real voice. This is most evident in
the indices that went back before World War II. For example, the Polity index rates
the United States a perfect democracy, a “10,” from 1871 on, somehow missing the
Jim Crow Laws altogether. Similarly, one would expect caste and the intimidation
of opposition voters at election time to be reflected in any widely useful definition of
democracy. Yet the procedural approach precludes recognizing such “social”
dimensions of political voice, and therefore sees no difference in the political rights
of Indians and Australians in the late nineteenth century.17 The current practice
tends to dichotomize too much between full democracy and autocracy, perhaps
because it has not sufficiently explored either pre-1950 history or the realities of
political voice in Latin American and Asia today.
The narrow procedural categorization of democracy will leave historians
unsatisfied. Most European countries extended the franchise in stages, “where
“Voice and Growth,” Page 7
Freedom slowly broadens down from precedent to precedent,” and a main task is to
sort out the causes and consequences of the transition from elite rule to democracy.18
In the history of the Americas the same difference between elite franchise and full
franchise is at the heart of differences in schooling, income inequality, and economic
growth. It seems to play a major role in explaining why the Southern states fell
behind in the United States, why Costa Rica is so far ahead of Guatemala and El
Salvador, and why North America pulled so far ahead of the rest of the
Hemisphere.19
The leading indices of political regimes also miss the gender dimension of
political voice. Even on strictly procedural grounds, one would want a concept of
political democracy to be affected by the denial or granting of voting rights to half
the adult population. Yet the historic shift to women’s voting rights clearly had no
effect on Banks’s regime categories or on the Polity index of democracy -- at least not
on the indices recorded for the United States, Canada, New Zealand, Norway, the
Netherlands, or Switzerland. What purpose were the indices of democracy and
freedom designed to serve, if denial of political voice to half the adult population
has no bearing?20
Finally, the current practice of classifying regimes focuses on the central
national government alone. This practice would do little damage in an historical
setting in which the key institutions are, and must be, centralized. Such was the case
centuries ago, when most damage to economic growth came from the top, so that
the Netherlands or England could shine just by having institutions that credibly
limited the power of the king or prince. But in a world where local governments
make the decisions that matter most to economic growth, the usual indices of
democracy or autocracy may fail to explain why different countries and regions
develop so differently. As we shall see, nineteenth-century Germany was one of the
settings in which the central regime is unrelated to local policies that drove much of
Germany’s growth.
“Voice and Growth,” Page 8
WHICH CHANNEL OF INFLUENCE -PROPERTY RIGHTS OR HUMAN CAPITAL POLICY?
The Property-Rights Channel
One of the keys to economic growth is surely freedom from confiscation of
one’s non-human capital. Scholars in economic history and political economy have
linked governance to economic growth through the kinds of institutions that affect
economic growth: property rights and the “rule of law.” What Montesquieu and
Adam Smith had intuited has become a well-articulated theory of institutional
history by Douglass North, Mancur Olson, Barry Weingast, and their co-authors. As
soon as either personal far-sightedness or revolutionary coercion makes some rulers
pre-commit to enforcement of laws limiting their powers to confiscate and tax
certain areas, those nodes of liberty develop cities, commerce, and industry.
Limiting the ruler’s power eventually strengthens the state itself.21
This explanation works especially well for earlier history, as one might have
surveyed it from, say, Scotland in 1776. So filled was the world with myopic and
rapacious rulers that any haven from confiscation could become a node of growth
and prosperity. The panel tests of DeLong and Shleifer, and of Acemoglu, Johnson,
and Robinson have shown that the property rights channel fits medieval and early
modern history systematically, not just in separate case studies.22
The same focus on the rule of law and property protection has now rightfully
become a prescription often applied by international agencies monitoring the
Second and Third Worlds.23 It is also the basis for repeated use of investor-service
indicies of “country risk,” which measure corruption, confiscation, and default all
over the world as indicators of property protection.
The Human Capital Policy Channel
“Voice and Growth,” Page 9
Yet as the economy evolves, different productive inputs become more crucial.
We have long known that economic development makes non-human property,
especially land, less and less important to growth, while human capital becomes
more important. The gradual shift toward an economy based on human knowledge
and communication shifts the mix of assets that democracies and autocracies need to
develop and protect. But sticking with the property rights of non-human capital,
many scholars have under-emphasized the human-capital channel, the one rightly
stressed in the EHA Presidential Addresses of Richard Easterlin and Claudia
Goldin.24 The evidence for this re-emphasis is compelling, and it means that advice
to today’s transition economies and developing countries needs to place at least as
much emphasis on policies toward education and health as on the protection of nonhuman property.
The comparative growth literature assigns a strong role to educational
attainment, even when the authors choose to downplay it. The Barro-Lee studies
end up featuring educational attainment as a key determinant of economic growth.25
The Clague-Keefer-Knack-Olson studies, while intent on showing the property
rights channel, nonetheless show a significant influence of schooling on GDP per
capita whenever they give it a chance.26 And although Bils and Klenow have
established that difference in schooling did not explain more than a third of growth
differences in a narrow accounting sense, they nonetheless show the likelihood that
there are at least some externalities from schooling that should supplement their
measure of the growth influence of schooling.27 The latest overall surveys of the
issue still find a strong effect of schooling on growth.28
That political voice strongly influences schooling and therefore growth is
increasingly clear in recent comparative work by economic historians. Historical
differences in the spread of suffrage go a long way toward explaining which
countries’ children got educated in the nineteenth and twentieth centuries.29
Suffrage also had a systematic effect on the tax effort put into education finance
“Voice and Growth,” Page 10
between the 1880s and the 1930s, a half-century in which some countries spread
voting rights down to the lowest-income classes, while others either remained
autocratic or kept political voice restricted to the top-income elite. The willingness
to spend tax money on primary education was significantly greater in full
democracies than in elite-vote limited democracies. In fact, the countries least
willing to spend taxes on mass primary education were not the autocracies, but the
elite-vote countries. The same was not true of public spending on university
education, however, for which elite-vote countries and autocracies were about as
willing to spend as were the full democracies.30
What makes the link between broad suffrage and primary schooling so
important is that the marginal growth effect of primary schooling is particularly
high in less developed settings, such as the pre-1914 history of the now
industrialized countries or the Third World today. In a well-governed country, the
true social returns to schooling should be equated across levels of schooling, and
equated with other rates of return throughout the economy. Failure on this front
explains much of the underdevelopment observed both today and in the past.
We have strong indirect evidence that poor societies systematically underfinance primary education. The evidence is indirect because it is confined to socalled “social rates of return” on the attainment of a higher level of schooling. These
rates of return are as encompassing as they can be, but some of the returns to
education are still left out. For one thing, such rates of return can only capture the
returns and costs of extra school years, not the returns and costs of raising the
quality of schooling at each level. That is, they can show only the damage done by
rationing schooling, not the damage from poor schooling. For another, they cannot
measure the net external or inter-generational benefits of education, and are “social”
only in that they include the public-budget effects of public financing and later tax
collection from more educated adults. For what they are worth, however, those
studies consistently show that the social rate of return on the extra (unattained)
primary schooling is much higher in today’s Third World than either the marginal
“Voice and Growth,” Page 11
returns on higher education in the same countries or the rates of return measured for
any level of schooling in high-income countries.31
Under-investment in primary schooling reflects two defects at once. First, it
reflects the usual imperfections of capital markets, which block low-income families
from borrowing to educate their children, whose high later incomes could have
repaid a loan at low prime interest rates. Second, it also reflects insufficient use of
taxpayer funds to conquer this capital market imperfection. Given the
pervasiveness of capital market imperfections and of external benefits from
education, taxpayer effort on behalf of public education has been key to raising
educational performance. For two centuries now, the global leaders in educational
attainment, test scores, and human earning power have been countries that have
relied on public funding at the primary and secondary levels. Tax money does not
simply displace private or philanthropic funding.32
The failure to equilibrate social rates of return suggests an elitist policy bias,
one that sacrifices GDP growth and discriminates against those who would benefit
from extra primary education -- particularly the poor, the rural, and females.
Some simple indicators can reveal an elitist bias in a country’s educational
policy, even without sufficient data to estimate rates of return. Compare that
country’s public-education expenditure and admissions patterns with those of highincome high-technology countries in the same era. The first fingerprint that an
elitist bias would leave relates to the primary school “support ratio”
Primary-school support ratio =
(public funding for primary school per child of primary-school age)
divided by (GDP per capita).
Note that the school-age population here is an entire age group, not just pupils, in
order to combine both support per pupil and the attendance or enrollment rate.33
Such a support ratio will typically rise with GDP per capita. A country’s
“Voice and Growth,” Page 12
educational policy leaves such an elitist fingerprint, Fingerprint #1, if it has a lower
support ratio for primary education than a typical country of the same income level,
or a poorer country, in the same historical time-period. In such a case, this
Fingerprint #1 means that the country is passing up some economic growth, either
to keep powerful groups from paying taxes or to keep the masses unschooled as an
object in itself. We will illustrate the use of this clue in the next section.
For the eighteenth and nineteenth centuries, this support ratio test is our best
prima facie clue to an elitist bias in educational policy, one that sacrifices some
economic growth. For the twentieth century, two other clues can support this one.
The twentieth century brought a general rise in public support for education at all
levels, in tandem with the shift toward greater and greater reliance on advances in
knowledge and skills. Thus for the twentieth century, elite bias can also show up as
relatively generous public funding for higher education, given that higher-income
and politically privileged families typically have better access to that higher
education. With taxpayers now subsidizing all levels of education and with greater
data availability, we can use two other clues that suggest elite bias at the expense of
overall GDP growth:
Relative support ratio for higher education =
(public support for tertiary education per pupil)
to (public support for primary education per child of primary-school age),
and
Inequality of support favoring the best-off =
A direct measure of the concentration of public support for education into
educating those with the highest levels of educational attainment, such as a
gini coefficient or a share of subsidies received by the best-educated ten
percent.
“Voice and Growth,” Page 13
Fingerprint #2 is left when policy gives a higher relative support ratio for higher
education than other countries with the same or higher average incomes.34 Granted,
it is conceivable that a poorer country might need to concentrate its education
budget on training at the top, so that national leaders and teachers are trained first,
before advances in schooling can trickle down to the masses. But the rate-of-return
evidence, plus smoking-gun historical narratives of elite antipathy to mass school
for its own sake, suggests a growth-sacrificing elite bias if Fingerprint #2 is found.
Similarly, Fingerprint #3 show up whenever a direct measure of inequality of public
funding favors the best educated groups, relative to typical practice in leading
countries. The calculated social rates of return are lower for tertiary education than
for primary, and there is no clear externality argument in favor of subsidizing higher
education more than primary education in a lower-income setting.
Aided by these clues to growth, we now turn to reinterpreting growth effects
of political voice, and the relative importance of the property rights and human
capital channels in conveying those effects. I have selected five settings in which
prevailing opinion seems to have underestimated either the case for democracy, or
the damage done by elite rule on the education front, or both.
FIVE REINTERPRETATIONS
Britain as Leader and Laggard
When did Britain lead the world in democracy, and through which channel
did it stimulate growth? The conventional answer is that from 1688 or earlier until
World War I British democracy shone because it supported private property rights,
and because its landed elite was willing to accommodate industrialization. While
the conventional stories are partly correct for the early modern era, the property
“Voice and Growth,” Page 14
rights story has been undergoing revision, while the downside of undemocratic
Britain’s fiscal and educational policy needs more attention.
The conventional tale is that Britain took off after 1688 thanks to the Glorious
Revolution’s putting constitutional limits on the throne’s ability to cheat creditors
and impose sudden taxes. Once Parliament became the supreme budgetary
authority, private businesses and holders of government bonds both felt more
protected. Private capital accumulation blossomed, and the state was able to fight
wars more effectively because it was more credit-worthy and levied more
predictable and acceptable taxes.35
Subsequent scholarship has reinforced the variant relating to government
creditworthiness, but not the role of 1688 in supplying cheaper loans for private
capital formation. Gregory Clark finds that the Glorious Revolution had no visible
effect on private interest rates, which were trending slowly downward long before
and long after 1688. On the other hand, he does find that 1688 lowered the interest
rates on the public debt of the newly constrained government. Unable to repeat
default episodes like the 1672 Stop of the Exchequer, the throne credibly precommitted to loan repayments, cutting the interest rate it was charged.36
If the only clear improvement in property rights springing from 1688 was an
improvement in royal creditworthiness and perhaps in the efficiency of British
taxation, what are we to do with the argument that the Revolution in 1688 protected
creditors lending to Britain’s private industrialists and sparked the Industrial
Revolution? The issue needs further exploration. We must consider the possibility
that improvements in English private capital supply were a much more gradual
affair, extending back to the middle ages, and improving slowly ever since. It
appears that real interest rates were declining persistently, if not monotonically,
from the 1251-1350 century to around 1900, not only in England but also in France,
Flanders, Germany, Netherlands, and Italy. Perhaps the story of improving capital
supply, presumably helped by improvements in private property protection, is a
“Voice and Growth,” Page 15
story to be told for much of Western Europe over six centuries, not a tale of a few
constitutional triumphs.37
Against the gains in British property rights and public finance we must weigh
the long-term costs of the bias of British politics toward the landed elite in the
eighteenth and nineteenth centuries. Granted, there are ways in which the landed
interest made its accommodation with liberalism and industrialization, helped by its
being a heavy urban landowner. In the final telling, we will still agree with the
judgment that the landed elite made this accommodation more smoothly in Britain
than elsewhere.38 Still, scholars have under-emphasized several ways in which it
actually checked Britain’s industrialization and growth. Perhaps with taxation as
with early British elite democracy, we tend to give early British fiscal institutions too
much credit because other countries’ institutions were even worse in those days.
The anti-growth side of landed elite power left some of its fingerprints on
Britain’s fiscal structure. The usual revenue calculations overlook ways in which
Britain taxed the advancing modern sectors. Any telling of the story of secure
British property rights should factor in the heavy stamp duties on commercial, legal,
and financial services. By the period 1815-1841 these taxes on such documents as
property and contract records had risen to claim 12-15 percent of government
revenue, more than the taxes on land and income. Any story of elite
accommodation to industrial capital and to the spread of knowledge must also
confront the duties on glass, windows, bricks, paper, and candles as well as those
stamp duties.39 We must also remember that customs duties on imports, which
were between 22 and 40 percent of government revenue throughout the period 17001870, discouraged exports of manufactures, and therefore manufacturing
production, for the simple reason that in the long run any tax on foreign trade
restricts both imports and exports.
The Corn Laws made that tax system even more regressive and anti-industrial.
Using simple “flypaper” calculations of tax incidence might seem to show that the tax
revenues took a bigger share of consumption from the landed rich than from others.40
“Voice and Growth,” Page 16
Yet the usual calculations of the tax revenues paid directly by the different classes miss
a major fiscal redistribution from workers and the poor, and from industrialists,
toward the landed interest. Britain’s Corn Laws raised the price of grain and basic
foodstuffs, drawing resources back into agriculture at the expense of industry -- yet
they are almost invisible in the government revenue accounts because they were
designed to be virtually prohibitive in the 1765-1792 and 1815-1843 eras. To make
wheat expensive was also to make bread expensive, raising the cost of living for the
landless masses. Statistical regressions suggest that the Corn Laws raised workers’
living costs by seven percent in the 1770s-1780s, and by 8-14 percent in 1820-1845. Any
reckoning of the fiscal policies of Georgian and early Victorian England must consider
these long episodes of policy-induced food scarcity, hurting both industry and the
poor.41 The only interlude of truly progressive taxation in the era of landed dominance
was the French War era 1793-1815, which included a temporary income tax. But as
soon as the war was over, this was repealed and the records were burned.
Turning to the education channel, some might expect that Britain fits the
current theme well because Britain was a leader in both democracy and schooling.
Britain does indeed fit the theme well, but on an opposing note: Britain was a
laggard, both in democracy and in schooling. In terms of democracy, it was not until
the 1880s that half of the adult male population could vote, a lag of decades behind
North America and France.42 In terms of schooling, both the enrollment figures and
the expenditure figures imply that Britain fell behind by mid-century or earlier, and
did not begin to catch up to the leaders until after the Fees Act of 1891 provided the
taxpayer support that was already provided in other countries.43
The public expenditure figures supply us with that “Fingerprint #1” of elite
bias in education policy, as shown in Figure 1. The lag of Britain in public support
for primary schooling can be seen in Figure 1 by comparing the support ratios on the
vertical axis either for the same years or for the same Maddison-estimated level of
real GDP per capita. As the fingerprint test requires, British taxpayers supported
primary education less than did some countries with lower average incomes in the
“Voice and Growth,” Page 17
same historical era. While the underlying political explanation is complicated, with
the education issue getting mixed in with religious and constitutional issues, the
nineteenth-century debate identifies aristocratic Tories, the Church of England, and
the House of Lords as groups consistently reluctant to have taxes paid to give the
broad mass of children a competitive secular education.44 By holding back public
mass education, Britain’s still-elite franchise in the early and mid-nineteenth century
held back Britain’s relative skills and GDP per capita for a few decades.
[Figure 1 about here.]
Thus the early lead and later lag in British political freedoms seem to have
altered the timing of British superiority, pointing away from the eighteenth and
nineteenth centuries. Britain led in freedom and living standards before 1688, along
with the Netherlands. Britain began slipping in ways better traced back into the
nineteenth century, not concentrated in the twentieth. We are better prepared for
these reinterpretations of Britain’s glorious era if we have studied the changes in
economic historians’ best estimates of Britain’s growth rates in terms of GDP per
capita. The United Kingdom’s growth rate per capita fell behind that of the
Americas and Australasia already in the 1820-1870 period, and behind most of
Europe as well during the 1870-1913 period.45 Furthermore, the estimates by Crafts
and Harley and by Clark show even lower rates of growth between 1700 and 1830.
Britain was already more advanced by 1700 than we had realized. British
institutions shone best in the earlier era up to the eighteenth century, when the rest
of the world was still generally mis-governed. Elite rule had its anti-growth side in
the eighteenth and nineteenth centuries.
The Germanies
“Voice and Growth,” Page 18
How does the delayed rise of Germany as an economic power illuminate the
relationship of political voice to economic growth? We are usually given two
different stories, one for Germany before Napoleon invaded, and one about the
nineteenth-century rise of Prussia and unified Germany. Neither story deals
effectively with the role of private property rights or of education.
The diversity of the Germanies has posed serious problems for the
interpretation of the growth effects of German governance. Was the political
fragmentation of Germany before the nineteenth century good or bad for economic
growth? One might have imagined that small principalities would compete as tax
havens, just as cities did elsewhere in Europe. Yet the pioneering attempts at
comparative classification of regimes and property rights up to the early nineteenth
century have retreated to treating all of Germany as an area uniformly ruled by
unconstrained rulers, with no protection for private capital, as if Germany were no
different from Albania or Sicily. Germany thus becomes a tale of poor growth under
local despots until the early nineteenth century.46
What happened next to governance and growth, during the nineteenthcentury rise of Prussia and unified Germany? A different literature takes over,
telling us that equally absolutist monarchs promoted economic growth from above,
without any new constitution constraining their power. How could the absolutism
switch from being bad to being good for private capitalism and economic growth?
And how was the strong growth achieved? The Gerschenkronian story of state and
bank control is inconsistent with Germany’s never having achieved as cheap a
supply of capital as Britain or France.47
To get the effect of German governance right, and to choose the right
institutional channel for its growth effects, we must first remember how Germany
excelled in the nineteenth century. The one clearly outstanding source of growth for
which the world envied nineteenth-century Prussia and Germany was its pioneering
system of mass education. Unfortunately, this source has also been mishandled.
Seeing an autocratic monarchy and a world-leading educational system, scholars
“Voice and Growth,” Page 19
have linked the two in an unsatisfactory ad hoc manner. Falling into a trap set by
this simple correlation and by the overuse of statutory history, they have argued
that the imperial state-building of the Prussian and German empires fostered
education because, well, it would make the state stronger -- leaving us to wonder
why the Austrian, Russian, Turkish and other empires concluded that their
ambitions required suppressing mass education.48
Yet the “absolutist” central government in Prussia and Germany had yielded
control of educational finance and development to localities right after the
humiliating defeat by Napoleon.49 Control from above consisted of little more than
periodic conservative edicts about curriculum and patriotism, plus pay scales for
teachers. Schools were funded locally, by local appointees, with the result that taxbased primary education flourished where the local demand was already stronger in
the late eighteenth century, namely in cities and toward the north and west. Where
Junker power and state funds were most evident, in the rural east, the least school
funding and the lowest graduation rates obtained, despite an imperial policy of
wanting to Germanize the Poles in school. The pattern in the smaller non-Prussian
states was similar.
The most promising way to resolve the puzzles of governance and growth in
the Germanies is to recognize that limited local freedom promoted growth through
the education policy channel, starting in the urban north and west in the late
eighteenth century and early nineteenth. The most conspicuous source of Prussian
and German growth was the source that the undemocratic central governments did
not control. In this decentralization of control over the amount of schooling to
provide, the Germanies were like North America and unlike Britain, where
Parliament retained direct control of education throughout the nineteenth century.
India’s Backward Democracy
Above all, our understanding of the relationship of political voice to
economic growth must come to grips with the fact that the world’s largest
“Voice and Growth,” Page 20
democracy is one of the world’s poorest nations. Since independence, India has had
a higher democracy index than France. Why is India’s average income still so low,
even among developing countries?50
Adding to the puzzle is that fact that India’s property-rights institutions are
also not bad in investors’ eyes, relative to those of other developing countries.
Granted, on the corruption front, some of the international ratings put India worse
than the average developing country, and lately even worse than China. Yet even
without a good corruption score, India in the 1990s ranked above the average
developing country in overall “rule of law and business environment,” and ranked
even above the world average in the categories “law and order,” “property rights
and rule-based governance,” and some government-efficiency indicators.51 If
democratic India’s general business environment is so acceptable, we have even
more reason to wonder why India is so poor.
To make headway on this deep puzzle, let us begin with the 1990s evidence
that India policy undersupplies primary education while being much more generous
at higher levels. This will establish that the human capital channel is the main
conduit from institutional problems to low incomes in India’s case. The clarity of
these current symptoms will at least help us begin the search for historical sources.
At the start of the 1990s, as India was beginning to emerge as an exporter of
software and other highly skilled services, almost half of Indian adults -- 36 percent
of men and 61 percent of women -- were illiterate. A consensus of in-depth studies
has found a serious distortion of Indian public funds in favor of higher education at
the expense of mass primary education.52 For example, the World Bank in 1992 was
clear in its recommendations for Indian educational policy:
The aggregate level of public spending on education is probably adequate....
But some changes are called for in the allocation of those resources. In
particular, more spending should be allocated to primary education, mainly
to improve its ability to retain students.... The shift in funding in favor of
“Voice and Growth,” Page 21
primary education can be achieved by increasing the contribution of private
financing in higher education.... The structure of enrollments and financing
arrangements result in a distribution of public spending that is skewed
toward the privileged.53
India’s anomalous educational policy stands out in an overall Asian
perspective. Table 1 exposes all three of those elitist fingerprints in educational
policy. First, the support ratio for public primary education, that same measure
used in Figure 1 above, was lower for India than for any other of these ten Asian
nations except India’s immediate neighbors, Bangladesh and Pakistan. The next two
fingerprints, the ones showing the relative generosity of taxpayer support for higher
education, confirm that the problem is not just meagerness of public funds. Only
Bangladesh is more biased in favor of the highest educated on all counts. India’s
policies seem to tilt away from primary education, by any regional or global
standard. To these three fingerprints, one could add that the Indian teaching
profession at all levels has been dominated by males, again more so than in any
other non-Muslim Asian nation other than Nepal and Cambodia.
[Table 1 about here.]
India’s lack of commitment to primary education manifests itself in huge
class sizes, teacher absenteeism, and high drop-out rates. The problem is worst in
India’s “heartland” states, Bihar, Orissa, Uttar Pradesh, and (until the late 1990s)
Madhya Pradesh. In the village of Palanpur in western U.P. in 1983-4, a single
teacher was responsible for carrying out the national mandate to educate all children
of ages 6-10. There were 158 such children.
[T]he most notable feature of the village school is that it has more or less
ceased to function. The root of the problem is fairly obvious. The single
“Voice and Growth,” Page 22
teacher [upper-caste son of the village headman] has a ‘permanent’ post, and
his salary, which is quite high by local standards, is effectively unrelated to
his performance.... [H]e has little incentive to exert himself.
In 1983-4, the village teacher was taking full advantage of these
circumstances. More often than not, he did not even take the trouble of
coming to school at all. When he did, he would be accompanied by ten or
twelve children at most, mainly sons and daughters of his own close
relatives.... This did not prevent him from cheerfully entering 135 names in
the school enrollment register.54
It is also in these poorer heartland states that secondary and higher education
seems to have been supplied most abundantly in recent decades. In 1966, for
example, the Lucknow National Herald voiced its suspicions by noting that India’s
poorer heartland states had a higher share of high school students going on the
university than Britain, France, Japan, or India’s better-off southern states. One
underlying factor seems to have been the entrepreneurial opportunity to create new
high schools and colleges as a political base and cash cow for siphoning government
grants at low cost. Running a primary school, which was handed down to villagelevel panchayat rule in the 1950s, is less lucrative, though it is still an opportunity to
solidify a local partisan political base.55
How could decentralized democracy fail to reform such policy failures,
decade after decade? If everybody really had voice in India, one would have to
conclude that democracy indeed failed to promote growth through the education
channel. The only way to save Churchill’s defense of democracy from this
indictment would be to show that India is not the full democracy that its franchise
rate would suggest. That is, one would have to show that much of the Indian
population has been systematically excluded from having any real voice in tax and
educational policy. One should also show evidence that the lack of full democracy
was greater in those heartland states where the education policy failure was
greatest.
“Voice and Growth,” Page 23
Even though nobody has put the whole chain of argument together yet, there
is at least a prima facie case that political voice in India has been highly restricted and
disproportional, despite the holding of full-suffrage elections. As an invitation to
the fuller research effort this historical subject demands, my task is to note some
clues that political voice has always been denied to the lower income groups, castes,
and tribes, particularly in the heartland states, and that this appears linked to elitist
education policy.
The first clue is that the symptom itself, an educational system designed for
the elite, has been a feature of India’s history at least back to Thomas Babbington
Macauley’s infamous Minute on Indian Education in 1835.56 Granted, every
generation of British and Indian leaders in the twentieth century gave lip service to
free public education for all. In the transition to Indian provincial autonomy the
1930s and 1940s, most provinces passed compulsory education laws. But in the
absence of funding and enforcement, these were no more effective in India than in
any other polity where unfunded compulsion tried to precede the private demand
for mass schooling. The gap between grants per university student and subsidies
per primary student apparently even widened under provincial autonomy in the
1930s and 1940s. Gandhi and the Congress Party leadership continued the rhetoric,
but declined to provide the funds needed for the daunting task of conquering
illiteracy. Gandhi himself added to the problem by demanding that alcohol could
not be legal, and therefore not taxed for schools and other programs, and by
refusing to abandon his scheme for ‘self-supporting” education in which illiterate
children would learn all they needed to know by working at menial jobs.57
Both in the transition to independence and since 1947, political voice in India
was limited, first in law and then in practice. The differences in democracy among
Sri Lanka, India, and Pakistan were already evident in the 1930s. Britain gave Sri
Lanka universal adult suffrage in 1931, only a few years after the last restrictions on
women’s suffrage were removed in Britain itself. Provincial elections were held
under this new franchise in 1931 and 1936.58 In India, by contrast, the Montague“Voice and Growth,” Page 24
Chelmsford reforms approved by Parliament in 1919 extended the suffrage only to
include more property-taxpayers, persons with educational qualifications, and
landholders. The landless and urban workers were still not included; in most
municipal areas the electorate was about 14 percent, and in rural areas it remained a
tiny 3.6 percent.59 For its part, the Muslim League wanted little to do with
democracy.
The differences in franchise and voting persisted into the Independence era.
Voter turnout in Sri Lanka rose from 55.8 percent of the electorate in 1947 to 76-78
percent in two elections of 1960, to 86.7 percent in 1977. By contrast, in India it rose
only from 46.6 percent in 1952 to 60.5 percent in 1977, and dropped back to 57.0
percent in 1980, even though the legal franchise share had risen from 55 percent to
99 percent across the 1960s and 1970s, and Pakistan has remained autocratic.60
These differences correlate with Sri Lanka’s much better performance in primary
education, though not in higher education, than either India or Pakistan. And
within India, the voter turnout rate again correlates with the relative development of
primary education and average incomes. Voting, primary schooling, literacy, and
income all continue to be higher around the rim, in the South, Punjab and Haryana,
and lower in the heartland states of Bihar, Madhya Pradesh, Rajasthan, and Orissa.
What mechanism might have linked limited political voice with the
discouragement of primary education? We know that single-member pluralities,
like the electoral institutions of India and the United States, create a bias in favor of
the largest and longest-organized political parties. The Congress Party was given
decades of clear primacy among political parties during its leadership of the
Independence movement. In the first thirty years of Independence its leadership
was hard to dislodge, and it won a majority of seats despite never capturing a
majority of votes.61 Once Congress’s educational policy had set the favoritism for
higher educational into the five-year plans, no lower-class or lower-caste opposition
could easily overturn that policy. Voice was effectively restricted by history and by
political institutions. One could view India under the “Congress Raj” as a case of
“Voice and Growth,” Page 25
Mancur Olson’s institutional arteriosclerosis. Political elites became increasingly
entrenched, and institutions were frozen in practice. In India’s case, that transition
may have secured the power not only of the well-off in the heartland states, but also
of teachers as a tenured lobby against parental voice, competition, and reform.
Yet surely a bedrock of political exclusion in India has been its tradition of
caste, tribe, class, and ethnicity. No matter how full the franchise or how much
power has devolved to provinces and to village panchayats, even the most local rule
seems to remain concentrated into long-organized groups. For its part, the central
government had tried to equalize power with affirmative actions giving the
“backward classes,” “scheduled castes” and tribes not only job quotas, but even
reserved legislative constituencies. Yet control over taxes and especially education
remains largely provincial, an arrangement that appears to have perpetuated the
handicap of primary schooling for the disadvantaged groups and the heartland
states.62 Tentatively, the answer might be that the world’s greatest democracy fell
behind because it was not much of a democracy in ways that were crucial for
education policy. In this respect, twentieth-century India may have been the mirror
image of nineteenth-century Germany: an ostensible democracy that failed to be
democratic on the education front, as opposed to an ostensible autocracy that led the
world in locally initiated education.
Asian Tigers, Crippled Dragon
Lee Kuan Yew is justly proud of Singapore’s growth and prosperity under his
rule. Singapore has developed both of the institutional channels featured here:
protection of business property rights and heavy subsidies to education. Lee has
extrapolated from his success to become a leading spokesman for autocracy in
troubled settings like postwar East Asia. For Lee, recent Asian history shows that
autocratic capitalism is far superior to democracy:
“Voice and Growth,” Page 26
“The regime in Beijing is more stable than any alternative government that
can be formed in China. Let us assume that the students had carried the day
in Tiananmen [Square] and they had formed a government... What kind of
China would they have today? Something worse than the Soviet Union.
China is a vast disparate country; there is no alternative to strong central
power.”63
But if Lee is right that autocracy works best within East Asia’s different
culture, it is well to compare all the experiences within that region. Is democracy a
drag on growth within East Asia, stretching from Mongolia to Indonesia, and from
Myanmar to Japan? Lee’s critics are correct. If one surveys performance from 1960
to 1998, the region’s overall experience shows a positive, not a negative, correlation
of democracy and economic growth. Omitting Japan or communist regimes does
not reverse this result.64 The only way that one can cast Asian autocracy in a good
economic light is to be extremely selective. Lee Kuan Yew has not seen fit to
comment on the smooth transitions to democracy in Korea and Taiwan. Basically,
his hypothesis is based on a self-congratulatory contrast of successful Singapore and
(pre-democratic) Taiwan with anybody growing more slowly under a different
regime.65
No defense of autocracy should be allowed to block our view of the greatest
peacetime policy disaster of all time, Mao Zedong’s Great Leap Forward. The best
estimates are that his policies of communization and rigid controls over food
distribution costs killed between 16 and 30 people between 1958 and 1961, more
than the best guess of about 7 million in the Soviet famine engineered by Stalin in
the early 1930s.66 These deaths and the related suffering, which were concentrated
in food-surplus areas, were definitely linked to the issue of democracy versus
autocracy. By 1958 serious criticism, or even scrutiny, of Mao’s policies was
effectively ended by the “Hundred Flowers” trap, in which he first called for
criticism from below and then persecuted those who gave it. No effective voices
“Voice and Growth,” Page 27
could challenge the Great Leap, even as its horror became obvious. Soon after the
famine, Chairman Mao himself, without referring explicitly to the famine, seemed to
draw a lesson about the need for “democratic centralism” when addressing about
7000 cadres:
“Without democracy, you have no understanding of what is happening down
below; the situation will be unclear; you will be unable to collect sufficient
opinions from all sides.”67
We must all strive to learn from Chairman Mao. He himself did not learn, however,
and soon imposed the Great Proletarian Cultural Revolution to purge critics, both
real and imagined. Education virtually halted for a few years. Amartya Sen has
plausibly drawn the larger lesson that major famines are unlikely to occur under
true democracies, when all adults and a free press are able to speak up.68 Having a
global market for food does no good if the autocrat blocks all access to it.
The frequent mishandling of the China case illustrates a larger drawback of
the recent scholarship on democracy and growth since 1950. Virtually all of it uses
as its sample a single cross-section of nations, or at best a time-shallow pool. This
makes it impossible to distinguish between the effects of any featured variable, say
democracy or property rights, and that dark unknown called “fixed country effects,”
a compendium of our ignorance about forces that may be unique to each nation.
Worse yet, the cross-sectional approach throws away the historical clues offered by
major regime changes in this last half-century. For China, as just noted, Mao and his
capitalist successors are lumped together. For Chile, scholars have thrown Frei,
Allende, Pinochet, and the democracy of the 1990s together, hoping to learn
something from the average of all regimes. For Korea, five major switches between
democracy and autocracy occurred since 1953, again obscuring the political meaning
of any overall average performance.
“Voice and Growth,” Page 28
The Welfare State: Not a Channel
Finally, how does the broader comparative history judge that third Churchill,
the young Liberal who wanted full democracy because it would redistribute from
landlords to pensioners and the workers? The conventional answer today is that full
democracy threatens to drag down economic growth through taxes and transfers
that stifle initiative.
What comparative quantitative history seems to show, however, is that
nothing flows through this channel. Even though fuller democracy probably raises
transfers,69 there is no clear effect of extra transfers on GDP, despite widespread
suspicions to the contrary. All kinds of empirical studies, as opposed to theoretical
modeling exercises and simulations, show that while excessive taxes and transfers
could lower GDP, they did not do so in the high-budget welfare states. I have called
this the “free-lunch puzzle” of the welfare state.
For this there seem to be two related reasons. First, the welfare-state
democracies have designed many of the transfers so as to promote health and to
minimize work disincentives. Second, the democracies that choose high-budget
welfare states have also chosen tax mixes that actually lighten the burden of average
and marginal rates on saving and investment for given average tax rates. Thus the
economic risks of redistribution were not manifested in the history of the
industrialized OECD countries.70
TOWARD BETTER TESTS: FITTING ECONOMETRICS TO HISTORY
We need not rely solely on such historical narratives to make the basic point
that recent statistical studies have missed the strength of the voice-growth link.
Even if we wear some of the blinders of the recent analyses confined to the data-set
club experiences, we get a more positive answer to Churchill’s question just by
“Voice and Growth,” Page 29
fashioning tests that get the feedbacks and the historical timing right. The few key
requirements are these:
(1) a panel of experiences in which the distribution of voice differed greatly
over both time and space;
(2) more information on the distribution of voice than the mere procedures of
executive and legislative power can reveal;
(3) careful thinking about the long lags that identify how political regime
history and GDP growth interact with each other; and
(4) special attention to human capital policies, especially public schooling.
A particularly good set of historical experiences for illuminating this whole
causal structure is era from the late eighteenth century to the early twentieth, in
which full-franchise democracy emerged in some countries, fail to emerge in others,
and died in still others. While the global sample since 1950 offers similar richness,
the earlier history gives us better data on a measurable dimension of political voice
not yet quantified for most of the postwar Third World, namely the share of adults
whose ballot box choices really mattered. A sample of 24 countries supplies the
necessary richness of experience, and of franchise and voting data for the
democracies among them, from the 1880s through the 1930s.71 It includes an even
split between democracies and autocracies, with movements in both directions
between these types of regimes.
The sample needs to include long lags. Political regimes are subject to
processes of selection and overthrow that respond to earlier political history and
current economic and social conditions. The growth of GDP, in the spirit of Barro
and other contributors to the econometrics of global growth, depends on earlier
backwardness of the country, and on the educational attainment of the adult labor
force, which is itself an accumulation over earlier decades. And government policy
toward education, which will affect later economic growth, is itself a response to
recent political and economic history. I have chosen a time span between
“Voice and Growth,” Page 30
observations that allows these forces to work themselves out. While an earlier paper
used decades, here I use eight-year intervals between observations. The
observations span as much history as the data frontier will allow. Thus the time
dimension of the current historical sample runs consists of eight benchmark years
spaced eight years apart: 1881, 1889, 1897, 1905, 1913, 1921, 1929, and 1937. This
particular choice allows us to view World War I as an event between benchmarks,
and also includes the fascist takeovers of Italy, Germany, and Austria.
This set of 192 historical experiences allows us to identify a causal structure
with three sets of equations.72
First, the political regime, represented by an autocracy index and the
franchise share among democracies, is a function of its own previous history over 24
years, and of recent levels of GDP per capita, urbanization, schooling, the state of the
global economy, and whether the country recently lost a war. The autocracy index
ranged from zero, for democracies and for benign autocracies such as Norway 18981913, to 9 or 10 for Mussolini’s Italy, Nazi Germany, and Thailand. Autocracy was
something close to a random walk, depending mainly on its own value eight years
earlier, though autocracy was also checked somewhat by the experience of fuller
democracy 24 years earlier. Among democracies, the extension of the franchise not
only depended on its own earlier history, but was also raised by higher recent
income levels (the Lipset effect), and by the embarrassment of recently losing a war,
as in France shortly after 1873 and Austria and Weimar Germany after World War I.
The rate of voter turnout behaved similarly to this share of the population
enfranchised to vote.
The education channel is charted by a second set of equations, sorting out the
determinants of public primary, public secondary, and total university enrollments
per child of the 5-14 age group. To judge the role of political voice, we must first
give other determinants of public enrollments their due.73 A higher level of GDP per
capita, lagged 8 years, clearly raises primary-school and total enrollments. A more
crowded school-age cohort, represented by the share of children 5-14 in the total
“Voice and Growth,” Page 31
population, depresses enrollment rates at all levels, presumably because it receives
less resources per child. A dominant Catholic Church cut primary school
enrollments by 419 students per 1000 children 5-14, other things equal, though it did
not cut secondary or university enrollments.
Four different kinds of political regime between the 1880s and the 1930s left
their contrasting fingerprints on education, fingerprints like those we have seen in
Figure 1 and Table 1. The four types to be distinguished here are firm autocracy
(e.g. Mussolini’s Italy, Nazi Germany, and Thailand), benign autocracy (e.g. Norway
1898-1913), elite democracy (e.g. Britain up to the 1880s), and full-voiced democracy
(Switzerland, North America, Australasia). Of these four, the two extreme groups
provided more primary education, other things equal. Both the firmer autocracies
and the full democracies committed more public resources to primary education
relative to the alliances of tame monarchs and elite classes in the other two settings.
Secondary and higher education were another matter. The full democracies tended
to provide fewer resources here, other things equal. Granted, twentieth century
North America led the way in public secondary education, but the regression results
imply that this is explained well enough by higher incomes and the lack of a
dominant religion.74
The overall effects of political regimes on growth flowed through all the
historic channels of institutional influence, not just the education channel. Once one
has given due credit to the state of the world economy and the country’s own degree
of backwardness, firm autocracy had a negative growth effect, even if it did not
retard education. Presumably the usual property-rights effects explain much of this
clear result. Among the other kinds of regimes, we find no clear net differences in
the growth effects of benign autocracies, elite democracies, and full democracies.
While their styles of education policy differed greatly, the net growth effect was
similar. All presumably benefited from their superior property rights, which full
democracies supplemented with more primary schooling and less higher education.
“Voice and Growth,” Page 32
CONCLUSION
All three Churchills were right about one thing, says the partial evidence
assembled here. The average democracy has been better for economic growth than
the average autocracy, at least in the formative years before World War II. Perhaps
Lee Kuan Yew is the exception that proves the bad-autocracy rule he tried to deny.
When it comes to wrecking economies, the tyranny of the majority over the voting
minority, popularized by our reading De Tocqueville, may have been no match for
the tyranny over those with no voice at all.
Among democracies, which of the three Churchills was right? Was the elitist
House of Lords democracy that Churchill the conservative defended in 1947
economically superior to the full-voice democracy endorsed by the other two
Churchills? As far as GDP per capita is concerned, they appear to have played to a
tie in those formative years before World War I. Yet their policy styles and their
distributive consequences differed. Compared to elite democracies, fuller franchises
delivered similar growth benefits, and probably greater equality, through primary
schooling and other public investments that the younger Churchill campaigned for
in 1909. Britain missed this opportunity in the nineteenth century, when mass
primary schooling was crucial, but Britain caught up in the twentieth.
As history and modernization evolve, the economic benefits of democracy
also evolve. Back in the seventeenth century, the net benefits might have centered
on private rights to non-human property, for which an elite democracy might have
sufficed. But as the importance of human skills and individual decision-making
discretion grows, autocracy may obsolesce, just as the gains from slavery
obsolesced.75 This likelihood was well expressed by a head of state who, like
Churchill, offered a summary defense of democracy when he was out of office.
Korea’s Kim Dae Jung, rebutting Lee Kuan Yew, sees a natural evolution toward
electronic democracy:
“Voice and Growth,” Page 33
The Asian economies are moving from a capital- and labor-intensive
industrial phase into an information- and technology-intensive one. Many
experts have acknowledged that this new economic world requires
guaranteed freedom of information and creativity. These things are possible
only in a democratic society. Thus Asia has no practical alternative to
democracy; it is a matter of survival in an age of intensifying competition.
The world economy’s changes have already meant a greater and easier flow
of information, which has helped Asia’s democratization process.”76
If Kim is right, the human-capital channel may have already become the most
important institutional link between political regimes and economic growth.
“Voice and Growth,” Page 34
REFERENCES
Acemoglu, Daron, and James A. Robinson. “Why Did the West Extend the
Franchise? Democracy, Inequality, and Growth in Historical Perspective.”
Quarterly Journal of Economics 115, 4 (November 2000): 1167-1200.
Acemoglu, Daron and James A. Robinson. “Reversal of Fortune: Geography and
Institutions in the Making of the Modern Income Distribution.” NBER
Working Paper 8460 (September 2001).
Acemoglu, Daron, Simon Johnson, and James A. Robinson. “The Rise of Europe:
Institutional Change and Economic Growth,” 28 March 2002 draft.
Banks, Arthur S. Cross-Polity Time-Series Data. Cambridge: MIT Press, 1971.
Barro, Robert J. “Democracy and Growth.” Journal of Economic Growth 1, 1 (1996): 127.
Barro, Robert J. Determinants of Economic Growth: A Cross-Country Empirical Study.
Cambridge: MIT Press, 1997.
Barro, Robert J. and Jong-Wha Lee. “International Comparisons of Educational
Attainment.” Journal of Monetary Economics 32, 3 (December 1993): 363-394.
Barro, Robert J. and Jong-Wha Lee. “Winners and Losers in Economic Growth.”
Proceedings of the World Bank Annual Conference on Development Economics.
Washington, World Bank, 1993, 267-314.
Bils, Mark and Peter J. Klenow. “Does Schooling Cause Growth?” American Economic
Review 90, 5 (December 2000): 1160-83.
Bollen, Kenneth A. “Political Democracy: Conceptual and Measurement Traps.”
Studies in Comparative International Development 25, 1 (spring 1990): 7-24.
Brewer, John. The Sinews of Power: Money, War, and the English State, 1688-1783.
Boston, Unwin, 1989.
Clague, Christopher, Philip Keefer, Stephen Knack, and Mancur Olson. “Property
and Contract Rights in Autocracies and Democracies.” Journal of Economic
Growth 1, 2 (June 1996): 243-276.
Clague, Christopher, Philip Keefer, Stephen Knack, and Mancur Olson. “Institutions
and Economic Performance: Property Rights and Contract Enforcement.” In
Christopher Clague (ed.), Institutions and Economic Development. Baltimore:
Johns Hopkins University Press 1997, 67-90.
Clague, Christopher, Philip Keefer, Stephen Knack, and Mancur Olson. “Democracy,
Autocracy, and the Institutions Supportive of Economic Growth.” In
Christopher Clague (ed.), Institutions and Economic Development. Baltimore:
Johns Hopkins University Press 1997, 91-120.
Clark, Gregory. “The Cost of Capital and Medieval Agricultural Technique.”
Explorations in Economic History 25, 3 (July 1988): 265-294.
Clark, Gregory. “The Political Foundations of Modern Economic Growth: England,
1540-1800,” Journal of Interdisciplinary History 26 (1996): 563-88.
“Voice and Growth,” Page 35
Clark, Gregory. “Land Hunger: Land as a Commodity and as a Status Good,
England, 1500-1910.” Explorations in Economic History 35, 1 (January 1998): 5982.
DeLong, J. Bradford and Andrei Shleifer. “Princes and Merchants: City Growth
Before the Industrial Revolution.” Journal of Law and Economics 36 (October
1993): 671-702.
Drèze, Jean and Amartya Sen (eds.). The Political Economy of Hunger. Three volumes.
Oxford; Clarendon Press, 1991.
Drèze, Jean and Amartya Sen. India: Economic Development and Social Opportunity.
New Delhi: Oxford University Press, 1995.
Easterlin, Richard A. “Why Isn’t the Whole World Developed?” This Journal 41, 1
(March 1981): 1-17.
Engerman, Stanley L, Elisa Mariscal. and Kenneth L. Sokoloff. “Schooling, Suffrage,
and the Persistence of Inequality in the Americas, 1800-1945.” Manuscript,
October 1998.
Engerman, Stanley L. and Kenneth L. Sokoloff. “The Evolution of Suffrage
Institutions in the New World.” NBER Working Paper 8512 (October 2001).
Fenoaltea, Stefano. “Slavery and Supervision in Comparative Perspective: A
Model,” this Journal, 44, 3 (September 1984): 635-668.
Ferguson, Niall. The Cash Nexus: Money and Power in the Modern World, 1700-2000.
New York: Basic Books 2001.
Flora, Peter et al. State, Economy and Society in Western Europe, 1815-1975. Frankfurt:
Campus Verlag, 1983.
Fohlin, Caroline. Financial System Design and Industrial Development: International
Patterns in Historical Perspective. Cambridge: Cambridge University Press,
forthcoming.
Friedman, Milton. Capitalism and Freedom. Chicago: University of Chicago Press,
1962.
Gastil, Raymond Duncan. “The Comparative Survey of Freedom: Experiences and
Suggestions.” Studies in Comparative International Development 25, 1 (spring
1990): 25-50.
Gerber, James B. “Southern White Schooling, 1880-1940.” Doctoral dissertation,
University of California – Davis, 1986.
Goldin, Claudia. “The Human Capital Century and American Leadership: Virtues of
the Past.” This Journal 61, 2 (June 2001): 263-292.
Green, Andy. Education and State Formation: The Rise of Education Systems in England,
France and the USA. New York: St. Martin’s Press, 1990.
Gurr, Ted Robert and Keith Jaggers. Polity98 Project. Regime Characteristics 1800-1998.
Computer file. Ann Arbor: University of Michigan, 1999.
Gurr, Ted Robert, Keith Jaggers, and Will H. Moore. “The Transformation of the
Western State: The growth of Democracy, Autocracy, and State Power since
1800.” Studies in Comparative International Development 25, 1 (spring 1990): 73108.
“Voice and Growth,” Page 36
Hannah, Leslie. “The Economic Consequences of the State Ownership of Industry,
1945-1990.” In R.C. Floud and D.N. McCloskey (eds.), The Economic History of
Britain since 1700, second edition, volume 3. Cambridge: Cambridge
University Press 1994, 168-94.
Helliwell, John F. “Empirical Linkages between Democracy and Economic Growth.”
British Journal of Political Science 24, 2 (April 1994): 225-48.
Hirschman, Albert O. Exit, Voice, and Loyalty: Responses to Decline in Firms,
Organizations, and States. Cambridge: Harvard University Press, 1970.
Homer, Sidney and Richard Sylla. A History of Interest Rates, third edition. New
Brunswick: Rutgers University Press, 1991.
India, Ministry of Education. Report of the Education Commission 1964-66: Education
and National Development. Delhi: Government of India Press, 1966.
Inge, William Ralph. “Our Present Discontents (August 1919),” in his Outspoken
Essays: First Series. London: Longmans, Green & Co. 1994, 1-34.
Jackman, Robert W. and Ross A. Miller. Culture, Institutions, and Political Behavior.
forthcoming.
James, Robert Rhodes (ed.). Winston S. Churchill -- His Complete Speeches 1897 - 1963,
volume II. New York: Bowker, 1974.
Kearney, Robert N. “Sri Lanka.” In Myron Weiner and Ergun Özbudun (eds.),
Competitive Elections in Developing Countries. Durham: Duke University Press,
1987, 77-111.
Keefer, Philip and Stephen Knack. “Why Don’t Poor Countries Catch Up? A CrossNational Test of Institutional Explanations.” Economic Inquiry 35, 3 (July 1997):
590-602.
Keefer, Philip and Stephen Knack. “Polarization, Politics, and Property Rights: Links
between Inequality and Growth.” Public Choice, forthcoming.
Kim, Dae Jung. “A Response to Lee Kuan Yew: Is Culture Destiny? The Myth of
Asia’s Anti-Democratic Values,” Foreign Affairs 73, 6 (November/December
1994): 189-194.
Knack, Stephen and Philip Keefer. “Institutions and Economic Performance: CrossCountry Tests Using Alternative Institutional Measures.” Economics and
Politics 7, 3 (November 1995): 207-27.
Lanjouw, Peter and Nicholas Stern (eds.). Economic Development in Palanpur over Five
Decades. Oxford: Clarendon Press, 1998.
Lindert, Peter H. “Historical Patterns of Agricultural Policy.” In C. Peter Timmer
(ed.), Agriculture and the State. Ithaca: Cornell University Press, 1991, 29-83.
Lindert, Peter H. “The Rise of Social Spending, 1880-1930.” Explorations in Economic History
31, 1 (January 1994): 1-36.
Lindert, Peter H. “What Limits Social Spending?” Explorations in Economic History 31, 1
(January 1996): 1-36.
Lindert, Peter H. “Democracy, Decentralization, and the Rise of Mass Schooling
before 1914.” University of California - Davis, Agricultural History Center
Working Papers 104 and 105 (April 2001).
“Voice and Growth,” Page 37
Lindert, Peter H. Social Spending and Economic Growth since the Eighteenth Century.
Two volumes. New York: Cambridge University Press, forthcoming.
Mackie, Thomas T. and Richard Rose. The International Almanac of Electoral History Fully Revised Third Edition. London: Macmillan, 1991.
Maddison, Angus. The World Economy: A Millennial Perspective. Paris: OECD, 2001.
Margo, Robert A. Race and Schooling in the South, 1880-1950. Chicago: The University
of Chicago Press, 1990.
Mann, Michael. The Sources of Social Power, volume II. Cambridge: Cambridge
University Press, 1993.
Massie, Joseph. Calculation of the Present Taxes Yearly Paid by a Family of Each Rank,
Degree, or Class. Second edition. London: T. Payne, 1761.
Mehrotra, Santosh and Enrique Delamonica. Public Spending for the Poor. Basic
Services to Enhance Capabilities and Promote Growth. Submitted for publication.
Mehrotra, Santosh, P.R. Panchamukhi, Ranjana Srivastava, and Ravi Srivastava.
Uncaging the Tiger Economy: Financing Elementary Education in India. Submitted
for publication.
Melton, J. Van Horn. Absolutism and the Eighteenth-Century Origins of Compulsory
Schooling in Prussia and Austria. Cambridge: Cambridge University Press,
1988.
Montesquieu, Baron de (Charles de Secondat). The Spirit of Laws. Translated and
edited by Anne M. Cohler, Basia Carolyn Miller, and Harold Samuel Stone.
New York: Cambridge University Press, 1989.
Moore, Barrington. The Social Origins of Dictatorship and Democracy: Lord and Peasant
in the Making of the Modern World. Boston: Beacon Press, 1966.
Mullen, E.N. “Democracy, Economic Development, and Income Inequality.”
American Sociological Review 53, ___ (___ 1988): 50-68.
North, Douglass C. Structure and Change in Economic History. New York: W.W.
Norton, 1981.
North, Douglass C. Institutions, Institutional Change, and Economic Performance.
Cambridge: Cambridge University Press, 1990.
North, Douglass C. and Barry Weingast. “Constitutions and Commitment: The
Evolution of Institutions Governing Public choice in Seventeenth-Century
England,” this Journal, 1989.
Nurullah, Syed, and J.P. Naik. A Students’ History of Education in India (1800-1965).
Calcutta: Macmillan, 1964.
O’Brien, Patrick K. “The Political Economy of British Taxation, 1660-1815.” Economic
History Review, second series, 41, 1 (February 1988): 1-32.
OECD, Centre for Educational Research and Innovation. Education at a Glance: OECD
Indicators, 1992 and 2001 editions. Paris: OECD, 1992, 2001.
Offer, Avner. Property and Politics, 1879-1914. Cambridge: Cambridge University
Press, 1981.
Olson, Mancur. The Rise and Decline of Nations. New Haven: Yale University Press,
1982.
“Voice and Growth,” Page 38
Olson, Mancur. “Autocracy, Democracy, and Prosperity,” in Richard Zeckhauser
(ed.), Strategy and Choice (New York: M.I.T. Press 1991).
Przeworski, Adam and Fernando Limongi. “Political Regimes and Economic
Growth.” Journal of Economic Perspectives 7, 3 (Summer 1993): 51-69.
Przeworski, Adam, Michael E. Alvarez, José Antonio Cheibub, and Fernando
Limongi. Democracy and Development: Political Institutions and Well-Being in the
World, 1950-1990. Cambridge: Cambridge University Press, 2000.
Psacharapoulos, George and Maureen Woodhall. Education for Development: An
Analysis of ‘Investment’ Choices. Oxford: Oxford University Press, 1990.
PROBE Team. Public Report on Basic Education in India. New Delhi: Oxford University
Press, 1999.
Ransom, Roger and Richard Sutch. One Kind of Freedom. New York: Cambridge
University Press, 1977.
Ricks, Christopher (ed.). The Poems of Tennyson, volume 1. Berkeley: University of
California Press.
Robinson, James A. “Are Endowments Fate?” draft, 2001.
Rudolph, Susanne Hoeber and Lloyd I. Rudolph (eds.). Education and Politics in India:
Studies in Organization, Society, and Policy. Cambridge: Harvard University
Press, 1972.
Sanderson, Michael. Education and Economic Decline in Britain, 1870 to the 1990s.
Cambridge: Cambridge University Press, 1999.
Schram, Stuart. Mao Tse-Tung Unrehearsed -- Talks and Letters: 1956-71.
Harmondsworth, Middlesex: Penguin, 1974.
Schultz, Kenneth A. and Barry R. Weingast. “Limited Governments, Powerful
States.” In Randolph M. Siverson (ed.), Strategic Politicians, Institutions, and
Foreign Policy. Ann Arbor: University of Michigan Press 1998, 15-49.
Sen, Amartya. Development as Freedom. Oxford: Clarendon Press, 2000.
Sirowy, Larry and Alex Inkeles. “The Effects of Democracy on Economic Growth
and Inequality: A Review.” Studies in Comparative International Development 25,
1 (spring 1990): 126-57.
Smith, Adam. 1993 (1776). An Inquiry into the Nature and Causes of the Wealth of
Nations. Edited by Kathryn Sutherland. New York: Oxford University Press.
Tan, Jee-Peng and Alain Mingat. Education in Asia: A Comparative Study of cost and
Financing. Washington: The World Bank, 1992.
Temple. Jonathan. “The New Growth Evidence.” Journal of Economic Literature 37, 1
(March 1999): 112-156.
Temple. Jonathan. “Growth Effects of Education and Social Capital in the OECD
Countries.” OECD Economics Department Working Paper 263. October 2000.
Downloaded from http://www.oecd.org/eco/eco, 2001.
United Nations Educational, Scientific, and Cultural Organization (UNESCO). World
Education Report. New York: UNESCO, 1991-1998.
Weiner, Myron. “India.” In Myron Weiner and Ergun Özbudun (eds.), Competitive
Elections in Developing Countries. Durham: Duke University Press, 1987, 37-76.
“Voice and Growth,” Page 39
World Bank. Primary Education in India. Washington: World Bank, 1997.
World Bank. India: Reducing Poverty, Accelerating Development. New Delhi, Oxford
University Press, 2000.
World Bank. World Development Report 1997. New York: Oxford University Press,
1997.
World Bank. World Development Report 2000/2001. New York: Oxford University
Press, 2000.
Zakaria, Fareed, “Culture Is Destiny: A Conversation with Lee Kuan Yew,” Foreign
Affairs 73, 2 (March/April 1994), 109-126.
Zakaria, Fareed. “The Rise of Illiberal Democracy.” Foreign Affairs 76, 6
(November/December 1997): 23-43.
Zastoupil, Lynn, and Martin Moir. The Great Indian Education Debate: Documents
Relating to the Orientalist-Anglicist Controversy, 1781-1843. Richmond, Surrey:
Curzon, 1999.
“Voice and Growth,” Page 40
Table 1. Three Fingerprints of Elite Bias in Education Policy India versus Other Countries in the
mid-1980s
Fingerprint #1
(below-average
values suggest
elitist bias)
Fingerprint # 2
(above-average
values suggest
elitist bias)
Public tertiaryPublic primary educatio
n
expenditure
expenditures
per child of
Bangladesh
China
India
Indonesia
Korea, Repub. of
Malaysia
Nepal
Pakistan
Philippines
Singapore
Sri Lanka
Thailand
Papua New Guinea
Ten Asian nations
Japan, 1995
United
States
Fingerprint #3
(above-average
values suggest
elitist bias)
Mid-1980s
inequality
of public funds
among
students ranked by
per pupil
/
primary-school public pre-prim. + educational
attainment
age as a % of
primary expend.
GDP/capita
per child
For 10%
of
Gini bestprimary-school
mid-1980s
age
educated
3.4
83.3
.82
72.0
7.9
25.2
.44
31.0
5.4
36.8
.66
61.0
13.7
6.7
.27
21.0
12.7
5.6
.16
13.0
14.0
13.6
.38
32.0
7.0
35.5
.57
54.0
4.0
31.8
5.8
8.7
.19
14.0
8.4
7.7
6.2
13.4
.33
28.0
13.7
2.9
.33
23.0
19.8
53.0
.62
54.0
8.5
17.5
17.3
15.7
0.9
1.4
.43
Memorand
um:
GDP/capit
a
for 1985,
in 1990 $,
per
Maddison
577
1522
1079
1972
5670
4157
713
1400
1964
10896
2234
3054
3497a
36.3
15332
20717
“Voice and Growth,” Page 41
OECD average, 1988
17.3
2.0
Sources and notes to Table 1:
a = From Penn World Tables 6.0, not from
Maddison.
The sources are Unesco, World Education Report 1998, and Tan and Mingat,
Education in Asia.
The ten Asian nations averaged together are Bangladesh, China, India,
Indonesia, Korea,
Malaysia, Nepal, Philippines, Sri Lanka, and
Thailand.
for Fingerprint #3, a one-year profile is used to synthesize the whole
educational cycle.
The Unesco source, used here for Pakistan, Singapore, and Japan, allows the calculation of
the support
ratio through two different methods. They do not give the same answers,
however.
One possible source of discrepancy is the inclusion of pre-primary expenditures with the
primary
school estimates.
“Voice and Growth,” Page 42
“Voice and Growth,” Page 43
ENDNOTES
Parliamentary Debates (Hansard), fifth series, vol. 444, 11 November 1947, Columns
206-207, with emphasis added to highlight the best-known passage.
Note that Churchill was attributing the now-famous dictum to somebody else
(“it has been said”). His reference was probably to William Ralph Inge, “Our
Present Discontents (August 1919),” in his Outspoken Essays: First Series (London:
Longmans, Green & Co., 1919, p. 5), who said: “Democracy is a form of government
which may be rationally defended, not as being good, but as being less bad than any
other.”
2 Only after 1958 did the Lords admit women, such as Lady Margaret Thatcher.
3 The performance record of the nationalized British Steel from 1968 to the late 1980s
was mixed. Its productivity advances were poor by most standards in the 1970s, but
very impressive in the 1980s. On the eve of privatization it had costs as low as any
in the world, though it is hard to know whether this was due to the looming threat
of privatization itself (Hannah, “State Ownership of Industry, 1945-1990,” 176-183).
4 James (ed.), Winston S. Churchill ... Complete Speeches, volume II, 1143 and 1382. For
the larger debate over taxing land during the People’s Budget campaign of 1909, see
Offer, Property and Politics, 242-253, 317-383; Daunton, Trusting Leviathan, 330-374;
and the sources cited there.
5 North, Institutions, Institutional Change and Economic Performance, 109-10.
6 Hirschman, Exit, Voice, and Loyalty, especially 30-54. Some have placed Milton
Friedman’s Capitalism and Freedom in the optimistic camp, but Friedman declined to
assert that democracy promoted capitalism. What he did assert is that capitalism
promoted freedom and democracy.
7 Gerschenkron, Economic Backwardness in Historical Perspective, was not explicit in
espousing autocracy, but its advantages underlie his argument and his choice of
historical case studies.
8 Each of these cases was a democracy in large part, if not a full one. In terms of the
Polity democracy index from 0 to 10, Weimar Germany, pre-Allende Chile, and
Ukraine 1991-95 were rated at 6, and Belarus was rated 7 and 8 in the 1991-94
period.
9 Zakaria, “The Rise of Illiberal Democracy.”
10 Przeworski et al., Democracy and Development, 178 and 271. Their study is the
deepest of the postwar statistical literature on democracy and growth, in that it
covers forty years, 137 countries, and the key issue of life expectancy missed by
other studies. Yet their impressive empirical offering has some avoidable
limitations. One is that their analysis has the limitation of sticking to a strict
dichotomy between autocracies and democracies. A second is that, like the rest of
the literature, they omit pre-1950 history and some disaster countries. Finally, as
noted below, their neutralist position on the growth issue seems to have blocked a
balanced presentation of the issue of effects on well-being.
1
“Voice and Growth,” Page 44
See, for example, Clague, Keefer, Knack, and Olson, “Property and Contract
Rights in Autocracies and Democracies,” and their “Democracy, Autocracy, and the
Institutions Supportive of Economic Growth.” This pair of studies finds that more
durable autocracies outperform more fragile ones. Their different results about the
effects of durable democracy are not consistent, however, and are clouded by the
difficulty of interpreting fixed-country effects in a cross-section.
12 Barro, Determinants of Economic Growth, especially xi, 58-60; Ferguson, The Cash
Nexus, Chapter 12. Barro suggests that the level of democracy attained by Malaysia
and Mexico might be about right, and that the democratizations by Chile, South
Korea, and Taiwan in the 1990s may have gone too far.
13 Przeworski and Limongi, “Political Regimes and Economic Growth;” and
Helliwell, “Empirical Linkages between Democracy and Economic Growth.” A
related review and critique of the pre-1990 literature is Sirowy and Inkeles, “The
Effects of Democracy.” The simultaneity and selectivity issues raised by the
Przeworski-Limongi critique in 1993 were later addressed in their (and co-authors’)
2000 book, with the neutralist conclusions mentioned in the preceding paragraph.
14 For example, America’s newly freed ex-slaves gave up a large share of the
potential income of women, children, and the elderly as soon as they were
emancipated. Ransom and Sutch, One Kind of Freedom, Chapter 3.
15 Good defenses of this strictly procedural definition in analyzing democracy and
growth are Knack and Keefer, “Cross-country Tests;” Clague, Keefer, Knack, and
Olson, “Property and Contract Rights in Autocracies and Democracies;” and
Przeworski et al., Democracy and Development, especially pages 33-36. Gurr and
Jaggers take a similar stand when defining “DEMOC” in the Polity data sets. So
does Kenneth Bollen (“Political Democracy”), though he also supports the counterargument I turn to in the next paragraph and footnote.
16 Kenneth Bollen’s critique of definitions of democracy stresses the same point: “Is
there no difference in the degree of political democracy, if 95 percent of men are
eligible in one country versus 20 percent in another?” (“Political Democracy,” 13.)
17 In the late nineteenth century, almost no Indians had a meaningful right to vote at
the regional or India-wide level. In the Australian colonies virtually all adult males
voted at the colony or territory level, and women could vote in Western Australia.
Yet the Banks and Polity data sets do not recognize either India or Australia as a
country with any political rights before Australia became a federated
Commonwealth in 1901 (Mackie and Rose, International Almanac, 1).
18 The spread of the European franchise is quantified by Flora et al, State, Economy
and Society in Western Europe, 1815-1975. The possible dynamic sources of this
spread are modeled in Acemoglu and Robinson, “Why Did the West Extend the
Franchise?” The passage quoted here from Tennyson (Ricks (ed.), The Poems of
Tennyson, 530-1) was cited by Churchill in that same speech of 11 November 1947.
Both Tennyson and Churchill used it not as a call to extend the franchise, but as a
call to make that change more gradual. Writing the original in 1833 or 1834,
11
“Voice and Growth,” Page 45
Tennyson “feared the results of the political agitation which had led to the Reform
Bill, 1832.” (Ricks, Poems of Tennyson, 530).
19 See Engerman, Mariscal, and Sokoloff, “Schooling, Suffrage, and the Persistence
of Inequality in the Americas, 1800-1945;” Engerman and Sokoloff, “The Evolution
of Suffrage Institutions in the New World;” Lindert, “Democracy, Decentralization,
and the Rise of Mass Schooling before 1914;” and Robinson “Are Endowments
Fate?”
20 While these illustrations come from the Polity index, because its authors dared to
cover a long span of history, the same omission is evident in leading regime
taxonomies confined to the period since the 1950s. See Banks, Cross-Polity TimeSeries Data; Gastil, “The Comparative Survey of Freedom;;” and the online World
Bank set of political regime indicators.
21 Montesquieu, Spirit of the Laws, especially 10-1, 340-1; Smith, Wealth of Nations,
especially Books II.iii and III; North, Structure and Change in Economic History and his
Institutions ; North and Weingast, “Constitutions and Commitment:” Olson,
“Autocracy, Democracy, and Prosperity” and his Power and Prosperity: Schultz and
Weingast. “Limited Governments, Powerful States.”
22 DeLong and Shleifer, “Princes and Merchants;” Acemoglu, Johnson, and
Robinson, “The Rise of Europe.”
23 For example, the first recommendation of the World Bank’s World Development
Report 1997 on The State in a Changing World is that “Efforts to restart development in
countries with ineffective states must start with institutional arrangements that
foster responsiveness, accountability and the rule of law.” (Page 157.)
24 Easterlin, “Why Isn’t the Whole World Developed?” and Goldin, “The Human
Capital Century.”
25 See Barro, “Democracy and Growth;” Barro, Determinants of Economic Growth;
Barro and Lee, “International Comparisons of Educational Attainment; and Barro
and Lee, “Winners and Losers in Economic Growth.”
26 Thus Knack and Keefer, ““Institutions and Economic Performance,” and Keefer
and Knack, “Why Don’t Poor Countries Catch Up?” find strong growth effects of
schooling, but schooling is not considered in Clague, Keefer, Knack, and Olson,
“Property and Contract Rights in Autocracies and Democracies,” or in their
“Democracy, Autocracy, and the Institutions Supportive of Economic Growth.” In
their chapter on “Institutions and Economic Performance: Property Rights and
Contract Enforcement,” the four authors find schooling variable insignificant, but
this is probably because their dependent variable is non-human investment and
because they force two correlated measure of schooling to compete against each
other in the same regressions.
27 Bils and Klenow, “Does Schooling Cause Growth?” The Bils-Klenow approach is
also narrow in the sense that it omits intergenerational effects such as the wellknown effect of female schooling on fertility and learning in the next generation.
“Voice and Growth,” Page 46
Temple, “The New Growth Evidence” and “Growth Effects of Education and
Social Capital in the OECD Countries.”
29 Again see Engerman, Mariscal, and Sokoloff, “Schooling, Suffrage, and the
Persistence of Inequality in the Americas, 1800-1945.” Similarly, within the United
States, their disenfranchisement retarded schooling for blacks and for poor whites.
On the racial gap in votes and schools, see Margo, Race and Schooling in the South,
Chapters 2 and 3. On the correlation between planter power and the lower level of
Southern whites’ primary and secondary schooling, with no shortfall in university
education, see Gerber, “Southern White Schooling, 1880-1940.”
30 See the appendix.
31 Psacharapoulos and Woodhall, Education for Development, surveys the rate-ofreturn literature worldwide.
32 In 1998, for example, taxpayers paid for over three-quarters of primary and
secondary educational expenditures in every OECD country. The same does not
hold for tertiary education, of course. Private funds paid for over half of tertiary
education in Japan, Korea, and the United States, and almost a quarter of tertiary
funding for the OECD as a whole (OECD, Education at a Glance, 2001 edition, 94).
For the two-century survey of this issue, see Lindert, Social Spending and Economic
Growth, Chapters 5 and 6.
33 In defining the school-age population it is important to choose the same age range
for all countries, even though the number of years spent in primary school may
vary. Choosing almost any age group in the under-20 range gives the same
comparative results if applied consistently.
Instead of GDP per capita, a more appropriate denominator might be the
average income per adult, a measure of society’s effort to raise the knowledge of
individual children relative to society’s ability to pay. But to simplify, let us use
conventional GDP per capita here.
34 Note the difference in denominators: higher education per pupil, but primary
education per child of primary-school age. The intent here is to omit the universityage population not receiving higher education, to better isolate the rates of public
subsidy to the truly privileged, while weighing down the support measure for
primary education by including those who received none. In this way, we make the
ratio a stiffer test of true bias in favor of the privileged.
35 The best known presentation of this view is North and Weingast, “Constitutions
and Commitment.” The part of the story emphasizing the efficiency of eighteenthand nineteenth-century century Britain’s tax system is found in O’Brien, “Political
Economy of British Taxation, 1660-1815;” Brewer, Sinews of Power; and Schultz and
Weingast, “Limited Governments, Powerful States.”
36 Clark, “The Political Foundations of Modern Economic Growth: England, 15401800.” Faced with his finding no effect on private interest rates, one who believes
that 1688 stimulated capital formation would have to show that it raised the demand
for loanable funds as much as it improved the supply of funds.
28
“Voice and Growth,” Page 47
Homer and Sylla, History of Interest Rates, Chapters X, XI, and XIII; Clark, “The
Cost of Capital” and “Land Hunger.”
38 Moore, Social Origins of Dictatorship and Democracy, especially 29-39; and Mann,
Sources of Social Power, volume II, 92-136.
39 The excise duty on glass, for example, was extremely high, around twice the pretax cost of glass. The fact that the excise duties on glass, paper were accompanied by
import protection may have compensated producers, but not purchasers, of these
capital goods and knowledge goods. See Daunton, Trusting Leviathan, 32-8.
40 See Massie, Calculations of the Present Taxes; and O’Brien, “Political Economy of
British Taxation, 1660-1815.”
41 Lindert, “Historical Patterns of Agricultural Policy,” 47-50 and 61-2. Adam Smith
said the same about the early versions of the Corn Laws imposed by “our country
gentlemen:” “So very heavy a tax upon the first necessity of life, must either reduce
the subsistence of the labouring poor, or it must occasion some augmentation in
their pecuniary wages, [which industrialists must pay].” Wealth of Nations, 315-6.
42 The electorate numbers and voting numbers can be found in either Flora et al.,
State Economy, and Society; in Mackie and Rose, International Almanac; or in the latest
CD-ROM version of the Arthur S. Banks data set. The age-group populations by
sex can be found in some of the Brian Mitchell historical statistics volumes.
43 Michael Sanderson (Education and Economic Decline, 29) similarly sees a British lag
in education up to 1890, with a delayed cost in terms of overall growth: “1870-1890
was the dangerous period when we risked falling behind and it was the last phase
when we had modestly good growth rates (1.2 per cent GDP per man year)
compared with our competitors, yet with a poor educational system.”
44 See Lindert, “Democracy, Decentralization, and Mass Schooling” and Social
Spending and Economic Growth, Chapter 5. For the expenditure figures and the
difficult best guesses about enrollment figures for Britain, see the appendices of
either of these two writings, and the sources cited there.
45 Maddison, Millennial Perspective, 264-5, using the British series of Crafts and
Harley to 1820, then Deane, and then Feinstein.
46 DeLong and Shleifer, “Princes and Merchants,” Table 2; Acemoglu, Johnson, and
Robinson, “The Rise of Europe,” and the Appendix below.
47 More generally, the Gerschenkronian story of integrated banks accelerating
industrial growth from above is found wanting in Germany, Japan, and Italy by
Fohlin, Financial System Design and Industrial Development..
48 Melton, Absolutism and the Eighteenth-Century Origins of Compulsory Schooling in
Prussia and Austria; Green, Education and State Formation.
49 The argument in this paragraph and the next is developed more fully in Lindert,
“Democracy, Decentralization” and Social Spending and Economic Growth, Chapter 5.
50 The Polity 98 data set (described in Gurr and Jaggers, Regime Characteristics 18001998) gives independent India a democracy rating of 9 out of 10 for 1950-1975. When
Indira Gandhi suspended civil liberties during her Emergency of 1975-1977, the
37
“Voice and Growth,” Page 48
rating dropped to 7 in 1976-1977. Since then it has been an 8, yielding an average
democracy index of 8.47 for 1950-1998. By comparison, France had an average index
of 8.00 over the same period. The Gastil index of civil liberties puts India behind
France for the 1972-1989 period. In that same period, however, India’s Gastil index
was nearly as good as Spain’s, and better than that of Brazil, Mexico, or Thailand
(civlibb in the Barro 1994 data set, described in Gastil, “The Survey of Freedom”).
In terms of PPP-adjusted GDP per capita, India in 1999 ranked 153rd out of
205 nations, still poorer than Zimbabwe (World Bank, World Development Report
2000/2001, 574-575.
51 World Bank, India: Reducing Poverty, Annex Table 4.1).
52 Tan and Mingat, Education in Asia, 1992; Drèze and Sen, India: Economic
Development and Social Opportunity, 1995; World Bank, Primary Education in India,
1997; PROBE Team, Basic Education in India, 1999; World Bank, India: Reducing
Poverty, Accelerating Development, 2000; Mehrotra and Delamonica, Public Spending
for the Poor. Basic Services to Enhance Capabilities and Promote Growth, submitted;
Mehrotra et al., Uncaging the Tiger Economy: Financing Elementary Education in India,
submitted.
53 Tan and Mingat, Education in Asia, 144-5.
54 Drèze, Lanjouw, and Sharma, in Lanjouw and Stern, Palanpur, 182.
55 J.P. Naik, a leading investigator of Indian education, summarized the rise of rentcollecting entrepreneurship thus in 1967:
In Uttar Pradesh they have a phrase, ‘The congress has abolished the
Zamindari in land and has created a Zamindari in education.’ Such
Zamindari [landowners] are managers of colleges, who are well fed, well
clothed, and maintain the own cars, all on the profits from the institutions
which they run. It is now recognised that running an educational institution
can be an important means of economic and political power.
See Rudolph and Rudolph, Education and Politics in India, chapters on Faizabad
District (U.P.), Mysore state, and Rajasthan. The quotation in this footnote is from
Page 94.
56 Macauley did not equivocate about English superiority and the need to
concentrate on training rulers, not masses:
I have no knowledge of either Sanscrit or Arabic. But.... a single shelf of a
good European library [is] worth the whole native literature of India or
Arabic.... In India, English is the language spoken by the ruling class....
{I]t is impossible for us, with our limited means, to attempt to educate the
body of the people. We must at present do our best to form a class who may
be interpreters between us and the millions whom we govern -- a class of
“Voice and Growth,” Page 49
persons Indian in blood and colour, but English in tastes, in opinions, in
morals and in intellect.
(As reprinted in Zastoupil and Moir (eds.), The Great Indian Education Debate ... 17831843, 165-6, 169, and 171.)
57 Nurullah and Naik, Students’ History of Education in India, Chapter 10.
58 Kearney, “Sri Lanka,” Weiner and ¨Ozbudun, Competitive Elections, 80.
59 Weiner, “India” in Weiner and ¨Ozbudun, 41.
60 Kearney, Table 3.9, on 101; Weiner, 52.
61 Jackman and Miller, Culture, Institutions, and Political Behavior, Chapter 1.
62 For the most recent measurement of the educational gaps by caste, tribe, gender,
income class, and state, see World Bank, Primary Education in India, 112-141.
63 Fareed Zakaria, “A Conversation with Lee Kuan Yew,” 122-3.
64 This conclusion is supported by simple correlations between the Polity index of
democracy and the rate of growth of GDP per capita between 1960 and 1998. In a
sampling of twelve East Asian countries -- China, Indonesia, Japan, North and
south, Korea, Malaysia, Mongolia, Myanmar, Philippines, Singapore, Taiwan, and
Thailand -- the correlation between democracy and growth is 0.33. Throwing out
Japan (as a country already fully developed by 1960) actually raises the correlation
to 0.39. The same positive relationship is robust to any treatment of British Hong
Kong before 1997, or to the addition of the Indochina nations to the sample.
65 If Lee were to try crediting earlier wise autocracy with preparing Korea and
Taiwan for mature democracy, his argument for protecting autocrats would share a
flaw with the old argument for “infant industry” protection. Parasitic rentskimmers use that argument to defend their vested interests forever.
66 The early deaths in those four years claimed between 2 and 4 percent of the
population. The directly food-related policy blunders were augmented by Mao’s
wasteful campaign to produce steel at the village level, whatever the economic and
environmental cost.
67 Mao Zedong, “On Democratic Centralism: Talk to an Enlarged Central Work
Conference 30 January 1962,” in Schram (ed.), Mao Tse-Tung Unrehearsed, 146. This
passage was previously cited by Sen, Hunger and Public Action, 213n.
While they note this case and repeat part of the Mao speech, Przeworski et al.
seem to be at a loss to handle its meaning, musing only that “it is not apparent
whether this is an argument strictly about avoiding disasters or about average
performance.” (Democracy and Development, 144.) Surely the disasters are part of the
average, and surely death matters along with the growth rate of GDP per capita.
68 Sen, “Freedoms and Needs,” New Republic, January 10 and 17, 1994, 31-8,
especially 34. For his underlying analysis of twentieth-century famines, see Hunger
and Public Action, especially Pages 210-4 on China’s famine, and Drèze and Sen
(eds.), The Political Economy of Hunger.
“Voice and Growth,” Page 50
Formal tests for 21 nations between 1880 and 1930 show that a rise in the share of
men voting, from around 30 percent to 80 percent or higher, significantly raises
public pensions and total social spending (Lindert, “Rise of Social Spending” and
Social Spending and Economic Growth, Chapters 4, 7, and 16). While this data set had
to stop with 1930, it is very likely that the 1930-1950 rise of democracy in
Scandinavia, Austria, Germany, and Italy also raised transfers and taxes.
70 Lindert, “What Limits Social Spending?” and Social Spending and Economic Growth,
chapters 10-12 and 18, and the earlier studies cited there. The dangers of
redistribution are not to be dismissed altogether, however, as shown by Allende’s
disastrous assault on the rich in 1970-1973.
71 The 24 countries are Argentina, Australia, Austria, Belgium, Brazil, Canada,
Denmark, Finland, France, Germany, Greece, Italy, Japan, Mexico, Netherlands,
New Zealand, Norway, Portugal, Spain, Sweden, Switzerland, Siam/Thailand, the
United Kingdom, and the United States. The fact that territorial boundaries changed
across World War I (e.g. from the Austrian half of the Austro-Hungarian Empire to
Austria alone) should pose little problem here, given that the sample is intended to
capture political changes. The only likely violation of the usual statistical
assumptions comes from the fact that serial correlation behavior might not be
consistent if the geography of the country changed.
72 See the appendix tables.
73 For primary and secondary education, I focus on public enrollments only, and not
private enrolments, for two reasons. The first is that data on public enrollments are
more consistently and reliably reported than private enrollments. The second is that
this paper focuses on government policy institutions, and it is appropriate to judge
the whole effect of public effort on economic growth, with the partial substitution of
public for private schooling built into the reduced-form estimates. Little harm is
done here, since a rise in public expenditure or enrollment raises the total
expenditure or enrollment as well. There is not much “crowding out” of private
school effort, as I argued in “Democracy, Decentralization, and The Rise of Mass
Schooling.”
74 On America’s twentieth century leadership, again see Goldin, “The Human
Capital Century.” On all the patterns mentioned here, see Lindert, “Democracy,
Decentralization, and the Rise of Mass Schooling,” and the appendix.
75 On how the need for workers’ informed discretion contributed to the decline of
slavery, see Stefano Fenoaltea, “Slavery and Supervision.”
76 Kim Dae Jung, “A Response to Lee Kuan Yew,” 192-3. I am indebted to Kim Suk
Ho for this reference.
69
“Voice and Growth,” Page 51
APPENDIX
Regression Equations for Enrollments, Growth,
and Political Regimes, 1881 - 1937
Appendix Table A. Regression Equations for School Enrollment Rates
per 1000 Children 5-14 , 24 Countries in 1881 - 1937
Dependent variables:
Public-school enrollments per 1,000 children 5-14
(1) primary only (2) primary only
coeff. |t|
coeff. |t|
School-age (5-14) share of total pop.
-7.80
ln (GDP/capita), 8 years earlier
167.6
Religion
Catholic dominance
-418.7
Protestant dominance
-154.9
Political regime variables (see also "Effects" below):
Autocracy index (0-10)
2.57
Did women vote 8 years earlier?
-1.6
Franchised as a % of population over 20
-0.8
Franchised share, squared
0.060
Franchised share, cubed
-.0005
Constant term
-506.6
Allowing for 23 fixed country effects?
"R sq.," equation F-statistic
Mean of the dep. var., std. error of estim.
.574
550.42
-4.59 (1.98) a
117.3 (7.42) **
(6.46) **
(1.55)
(1.11)
(0.10)
(0.34)
(1.15)
(1.68) a
(2.87)
No
Effects of selected shifts toward more electoral democracy:
(a) from benign non-democracy to 30% franchis
16.6
(b) from 30% franchise to 80% franchise
48.2
(c) from benign non-democracy to 80% franchis
64.8
(d) from benign non-democracy to 100% franch
21.3
Type of equation
(2.94) **
(8.71) **
4.42
-3.4
-0.61
0.057
-.0005
-149.6
(2.48) *
(0.27)
(0.36)
(1.47)
(2.10) *
(1.00)
Yes
28.7
0.98
(0.51)
(1.49)
(3.21) **
(1.21)
pooled GLS
(** = significant at the 1% level, two-tail; * = significant at the 5% level;
a = significant at the 7% level; b = significant at the 10% level.)
.879
550.42
20.08
51.26
71.35
31.15
46.6
0.97
(0.85)
(2.11) *
(4.06) **
(1.92) a
pooled GLS
Appendix Table A, continued
Dependent variables:
Enrollments per 1,000 children 5-14
(3) primary plus (4) university
secondary (public) (public + private)
coeff. |t|
coeff. |t|
Primary enrollment rate, 8 years earlier
0.82
School-age (5-14) share of total pop.
-2.8
ln (GDP/capita), 8 years earlier
79.3
Political regime variables (see also "Effects" below):
Autocracy index (0-10)
1.2
Did women vote 8 years earlier?
-3.8
Franchised as a % of population over 20
-0.78
Franchised share, squared
0.030
Franchised share, cubed
-0.0002
Greece in the 1920s (secondary overcounte 101.1
Constant term
-421.3
Allowing for 23 fixed country effects?
"R sq.," equation F-statistic
Mean of the dep. var., std. error of estim.
Number of non-zero observations, out of 192
##### **
(1.48)
(6.78) **
-0.0018 (0.52)
-0.7 (4.21) **
6.6 (5.63) **
(0.80)
(0.31)
(0.39)
(0.63)
(0.77)
(2.83) **
(4.34)
-0.17
5.0
0.43
-0.011
0.00007
-5.6
-31.2
Yes
.983 338.0
581.8 0.97
.793
5.6 3.09
7
1.5
-2.3
-0.8
-0.2
Type of equation
tobit
(** = significant at the 1% level, two-tail; * = significant at the 5% level;
a = significant at the 7% level; b = significant at the 10% level.)
**
**
**
**
*
Yes
Effects of selected shifts toward more electoral democracy:
(a) from benign non-democracy to 30% franchi
-2.6 (0.10)
(b) from 30% franchise to 80% franchise
18.8 (0.65)
(c) from benign non-democracy to 80% franchi
16.2 (0.99)
(d) from benign non-democracy to 100% franc
1.3 (0.11)
pooled GLS
(1.16)
(5.14)
(3.01)
(3.32)
(3.40)
(2.16)
#####
(2.46) *
(3.44) **
(2.21) *
(0.56)
Notes and sources to Table A:
The sample consists of 24 countries over eight benchmark years. The 24 countries are Argentina,
Australia, Austria, Belgium, Brazil, Canada, Denmark, Finland, France, Germany, Greece, Italy, Japan,
Mexico, Netherlands, New Zealand, Norway, Portugal, Spain, Sweden, Switzerland,
Siam/Thailand, the United Kingdom, and the United States. The eight benchmark years
are spaced eight years apart: 1881, 1889, 1897, 1905, 1913, 1921, 1929, and 1937.
The fact that territorial boundaries changed across World War I (e.g. from the Austrian half
of the Austro-Hungarian Empire to Austria alone) should pose no problem here, given that the
sample is intended to capture political changes. The only likely violation of the usual statistical
assumptions comes from the fact that serial correlation behavior might not be consistent if the
geography of the country changed.
The enrollment rates are from Lindert, "Democracy, Decentralization, and Mass Schooling before 1914,"
University of California - Davis, Working Papers 104 and 105 (April 2001), Appendix A. To
interpolate between my decadal benchmark estimates, I used some of the enrollment figures from
the Arthur S. Banks CD-ROM for 1815-1999. But in some cases, especially the UK, I prefer mr own
estimates over those than Banks presents without citing his sources.
The franchised are the shares of the over-20 population legally entitled to vote, in settings where
I judged the voting power to be real (see below). For years when women were not yet entitled to
vote, the over-20 population refers to men only. Alternative regressions used the actual voter
turnout instead of the franchise share used here. The results were qualitatively the same, both in the
regressions using voter turnout and in similar regressions on the 1880-1930 decadal sample results
reported in Agricultural History Center Working Papers 104 and 105. The franchise and voting
shares are from the Arthur S. Banks cross-polity CD-ROM for 1815-1999, which draws them
mainly from Mackie and Rose (1991). The autocracy index is from the Polity 98 version
the Gurr-Jaggers Polity data set.
The franchised voting power was judg ed to be illusory and not real in cases where the Banks indexes
and the Mackie description of franchise institutions suggested that voters had little power
over the legislature and the chief executive, despite their actually voting in legislative elections.
Thus I entered zeroes for the franchise in these cases where elections were actually held:
Belgium, Germany and Italy up to World War I; Norway to 1882, and Sweden to 1907.
Both the autocracy index and the franchised shares are predicted values, rather than actual
observed values. The instrumental-variable equations generating these predictions are
the political-regime equations in Appendix Table C.
Catholic dominance = the amount of the Catholic majority among religious declarants. It equals the
maximum of either 0 or the Catholic share minus .50. It takes on positive values for nine countries.
Protestant dominance = the corresponding majority margin for Protestant countries, with some
cases judged to involve no dominance despite a Protestant majority. It equals nearly 0.50 for
Denmark, Finland, Norway, and Sweden. It equals 0.16 for the UK before the separation of
Ireland, and 0.10 for 1921-1937.
The religion data are mostly from Annuaire Statistique de la France for the 1930s. Those from
France, the UK and a few other countries are from encyclopedias, in some cases for postwar years.
"Benign" non-democracy here refers to a polity with an autocracy rating of zero, but with enough
impediments to legislative effectiveness and enough power of the monarch for me to disregard
any suffrage rates, setting them at zero despite the occurrence of elections. The only pure example
in the sample is Norway 1898-1913, though prewar Belgium came close, with autocracy = 1.
For Greece in the 1920s, I used the Banks data series on secondary and higher education rather than
the less complete Mitchell series. However, the Banks series seems to overcount secondary
enrollments, partly at the expense of tertiary enrollments. This necessitated adding the
"Greece in the 1920s" variable to capture the temporary miscount.
The test statistics listed under "effects" at the bottom of the table start from the most limited
autocracies, those with a Polity AUTOC index of 0, combined with my judgment that they
were nonetheless not democracies. For stricter autocracies, note the autocracy index coefficient.
Regressions were run using the POOL command in SHAZAM 8.0, using the option that sets the
same first-order rho coefficient for all countries in making the Cochrane-Orcutt transformation.
Appendix Table B. Regression Equations for Eight-year Growth Rates
of GDP per Capita, 24 Countries in 1881 - 1937
(5) Growth of
GDP per capita
coeff.
Growth of the global economy
0.736
This country's backwardness, 8 years earlier
0.091
Agriculture's share of labor force, 8 yrs. earlier
-0.010
Public primary school enrollments, 8 years earlie #######
Public secondary school enrollments, 8 years earl 0.00053
University enrollments, 8 years earlier
-0.0027
Autocracy index (0-10)
-0.017
Franchised as a % of population over 20
-0.00017
Constant term
-0.028
Allowing for 23 fixed country effects?
|t|
#####
(3.73)
(0.17)
(1.69)
(1.92)
(1.64)
(4.20)
(0.69)
(0.95)
(6) Growth of
GDP per capita
coeff.
**
**
|t|
0.750 ##### **
0.092 (3.85) **
-0.023 (0.37)
b
a
**
No
-0.019 (4.90) **
0.000 (0.00)
0.021 (0.92)
No
"R sq.," equation F-statistic
Mean of the dep. variable, std. error of estimate
.579 33.0
0.097 0.986
.537 43.1
0.097 0.501
Type of equation
pooled GLS
pooled GLS
(7) Growth of
GDP per capita
(8) Growth of
GDP per capita
Growth of the global economy
This country's backwardness, 8 years earlier
Agriculture's share of labor force, 8 yrs. earlier
Public primary school enrollments, 8 years earlie
Public secondary school enrollments, 8 years earl
University enrollments, 8 years earlier
Autocracy index (0-10)
Franchised as a % of population over 20
Constant term
Allowing for 23 fixed country effects?
"R sq.," equation F-statistic
Mean of the dep. var., std. error of estim.
coeff.
|t|
0.724
0.331
0.129
0.00005
0.0019
-0.0084
-0.019
0.000
-0.203
#####
(6.66)
(1.41)
(0.77)
(4.79)
(3.65)
(4.30)
(0.19)
(2.51)
Yes
.730 16.7
0.097 0.984
coeff.
**
**
**
**
**
|t|
0.750 ##### **
0.263 (5.18) **
-0.0069 (0.08)
-0.021 (4.42) **
-0.00021 (0.82)
-0.077 (1.19)
Yes
.642 12.3
0.097 0.983
Notes and sources to Appendix Table B:
See also the notes and sources to Appendix Table A.
The "growth of the global economy" variable equals the log-growth, over the last eight years,
of the GDP per capita for eleven annual-data countries: Belgium Canada, Denmark, France,
Germany, Italy, Netherlands, Norway, Sweden, UK, and US, from Maddison,
Monitoring the World Economy.
The backwardness variable equals the log-difference between the GDP per capita
of the highest-income country (either Britain or America) and
the GDP per capita of this country, eight years earlier.
All educational enrollments are percentages of the same 5-14 age group population.
This is not the usual school-age population, especially for university students.
Yet the same denominator was used thhroughout in order to make the enrollment
figures, addable across levels of schooling. In any case, all young age-group
populations tend to move similarly as shares of total population.
Both the autocracy index and the franchised shares are predicted values, rather than actual
observed values. For the instrumental-variable equations generating these predictions,
see Appendix Table C.
Regressions were run using the POOL command in SHAZAM 8.0, using the option that sets the
same first-order rho coefficient for all countries in making the Cochrane-Orcutt transformation.
Appendix Table C. Regression Equations for Political Regimes:
the Autocracy Index and the Franchise Share,
24 Countries in 1881 - 1937
(9) The autocracy (10) The franchised
index (Polity)
share of adults
coeff.
The autocracy index, lagged 8 years
0.976
The autocracy index, lagged 16 years
-0.278
The autocracy index, lagged 24 years
0.254
Franchised as a % of adult population, lagged 8 yr 0.011
Franchised as a % of adult population, lagged 16 y 0.003
Franchised as a % of adult population, lagged 24 y -0.069
ln (GDP/capita), 8 years earlier
0.132
Growth of the global economy
3.654
Lost a war in the last 8 years
-17.154
Lost a war 9-16 years ago
-14.825
Share of population in cities > 50,000, 8 years earl 1.139
Primary + secondary enrollment rate, 8 years earli -0.0011
Constant term
-1.466
Allowing for 23 fixed country effects?
|t|
(6.25) **
(1.36)
(1.56)
(0.90)
(0.22)
(4.25) **
(0.16)
(1.42)
(0.01)
(0.01)
(0.25)
(0.82)
(0.25)
No
coeff.
|t|
1.10
0.38
-1.48
0.96
-0.13
0.12
31.58
-26.17
82.78
51.20
-67.59
0.029
-259.83
(0.53)
(0.14)
(0.71)
(6.54)
(0.74)
(0.78)
(2.89)
(0.94)
(3.44)
(2.40)
(1.78)
(1.61)
(3.25)
**
**
**
*
b
No
"R-squared"
Mean of the dep. variable, std. error of estimate
Number of non-zero observations, out of 192
.693
1.302 2.521
104
.763
26.86 30.779
98
Type of equation
Tobit
Tobit
Notes and sources to Appendix Table C:
See also the notes and sources to Appendix Tables A and B.
The variable "lost a war in the last decade" = 1 for France 1881, and for Germany and Austria
in 1921 and 1929, otherwise = 0.
The share of the population living in cities greater than 50,000 in population is from the CD-ROM
of the Arthur S. Banks data set for 1815-1999.
In the case of tobit regressions the "R-squared" parameter is the squared correlation between
observed and expected values, and the mean of the dependent variable is the expected
expected value at the mean values of all independent variables.
Regressions were run using the TOBIT command in SHAZAM 8 0